Akbar Aman

AI Intern | AHEAD

Duration: May 2026 - Present

Location: Chicago, IL (Hybrid)

Role Overview

Supporting AHEAD's internal enterprise AI team by helping turn platforms like Glean and Claude into usable, governed, and scalable internal workflows. The role sits between AI platform operations, knowledge systems, workflow design, agent and skill development support, QA, documentation, and internal enablement.

Focus Areas

• Enterprise AI Enablement: Supporting internal adoption of GPT and agentic platforms across presales, operations, and business functions
• Connector Quality & Governance: Assessing data connectors, validating permissions and crawl behavior, and documenting setup status, risks, and blockers
• Agent Testing: Creating test cases, validating expected behavior, logging QA findings, and identifying edge cases or missing instructions
• Skills Development: Drafting reusable skill definitions with clear instructions, triggers, guardrails, ownership, and documentation
• Knowledge Systems: Building SOPs, templates, request guides, decision logs, and Confluence/SharePoint structures that make AI work repeatable
• Workflow Standardization: Translating ambiguous internal AI use cases into reusable artifacts, documentation, and business-facing summaries

Typical Deliverables

• Connector assessments, validation notes, and quality audit findings
• Agent specs, QA logs, test cases, and improvement recommendations
• Reusable skill definitions, catalog entries, SOP pages, templates, and request guides
• Concise research summaries, slide decks, and governance recommendations for internal stakeholders

AI Safety Evaluator / Red Team Prompt Engineer | LinkedIn

Duration: Apr 2026 - July 2026

Location: Remote | Contract

Role Overview

Supported LinkedIn's AI safety initiatives through adversarial prompt generation, structured safety classification, severity scoring, and policy-grounded rationale writing.

Key Responsibilities

• Generated adversarial prompts to evaluate model robustness, safety boundaries, and reliability
• Applied structured safety classifications and severity scoring to model outputs
• Produced policy-grounded rationales to support consistent evaluation workflows
• Contributed to production-oriented LLM trustworthiness and safety evaluation processes

Quantitative Research & AI/ML Engineering Intern | Crypt0nest.io

Duration: Jan 2026 - May 2026

Location: Remote

Company: Tekly Studio - Crypt0nest Quantitative Research Lab

Role Overview

Working within Tekly Studio's Crypt0nest quantitative research lab on the development of AI-driven portfolio intelligence systems for crypto markets. Operating in a high-agency, production-oriented environment focused on reproducible research, clean system design, and translating machine learning theory into end-to-end applied infrastructure.

Key Responsibilities

• Data Pipelines: Contributing to the development and maintenance of robust data pipelines for crypto market data ingestion, processing, and storage
• Feature Engineering: Designing and implementing features for quantitative trading strategies with focus on signal quality and predictive power
• Machine Learning Workflows: Building and optimizing ML workflows used to replicate and validate quantitative trading strategies
• Backtesting & Validation: Working under strict backtesting and data integrity constraints to ensure strategy reliability and prevent overfitting
• Portfolio Intelligence Systems: Contributing to AI-driven portfolio intelligence systems that analyze and optimize crypto market positions
• Intern Lead Support: Promoted to Intern Lead for peer pull requests, technical questions, and code review support

Technical Focus Areas

• Quantitative Research: Replicating and validating trading strategies using rigorous statistical and mathematical methods
• AI/ML Infrastructure: Translating machine learning theory into production-ready, end-to-end applied infrastructure
• Reproducible Research: Implementing clean system design practices ensuring research reproducibility and code maintainability
• Data Integrity: Maintaining strict data quality standards and backtesting protocols to ensure strategy validity
• Crypto Markets: Deep engagement with cryptocurrency market dynamics, trading patterns, and portfolio optimization

Environment & Culture

High-agency, production-oriented environment where interns are trusted with meaningful contributions to real quantitative research systems. Emphasis on clean code, reproducible research methodologies, and bridging the gap between theoretical ML concepts and practical trading infrastructure.

Technical Stack

• Machine Learning: PyTorch, TensorFlow, scikit-learn for quantitative model development
• Data Engineering: Python, pandas, NumPy for data pipeline construction and feature engineering
• Quantitative Finance: Backtesting frameworks, statistical analysis, portfolio optimization techniques
• Infrastructure: Production ML systems, data integrity validation, reproducible research workflows
• Crypto Markets: Market data APIs, trading strategy validation, portfolio intelligence systems

Software Engineer | Arkboosted

Duration: Sep 2025 - Present

Location: Chicago, IL (Remote)

Company Website: arkboostedads.com

Role Overview

Owned end-to-end execution of secure, scalable web and AI systems for client organizations through subcontracting work. Responsibilities span system architecture, backend development, deployment, iteration, and maintenance with a security-first development approach across all projects.

Key Technical Achievements

• Production Websites & Landing Pages: Delivered secure, high-converting websites for client organizations with modern technologies and conversion optimization
• AI-Powered Support & Intake System: Designed and built a pgvector-backed RAG (Retrieval-Augmented Generation) agent serving a 700+ organization network to unify knowledge retrieval, automate intake processes, and streamline support escalation workflows
• Internal AI Tooling: Developed comprehensive audit automation platform using FastAPI and React that automates SEO analysis, security scanning, and performance testing—reducing manual audit time by approximately 90% and generating actionable reports
• Business Growth: Secured 3+ new clients in first month, driving 200-275% month-over-month revenue growth through automated business processes

Technical Stack & Capabilities

• Backend Development: FastAPI, Python, RESTful APIs, database design and optimization
• Frontend Development: React, JavaScript, responsive design, conversion-focused UI/UX
• AI/ML Systems: RAG architecture, pgvector integration, LLM workflows, knowledge retrieval systems
• Infrastructure: Secure deployment, scalable system architecture, performance optimization
• Security: Security-first development approach, vulnerability assessment, secure coding practices
• Automation: Process automation, audit tooling, workflow optimization

Core Services Delivered

• Web Development: End-to-end website development with security-by-design principles, modern tech stacks, and conversion optimization for client organizations
• AI Systems: Custom AI-powered solutions including RAG agents, knowledge management systems, and intelligent automation workflows
• Development Tooling: Internal tools for audit automation, performance analysis, and actionable reporting to streamline operations
• System Architecture: Scalable, secure system design and implementation for client needs

Technical Consultant | Fibinaci (formerly Nexus)

Duration: Dec 2025 - Jan 2026 (2 months)

Location: Princeton, New Jersey, United States (Remote)

Former Website: nexusdevops.com

About Fibinaci

Fibinaci is the renamed and refocused evolution of the former Nexus platform. The concept moved from a broad open-source developer platform toward an enterprise AI infrastructure direction: mapping company systems, data, permissions, and workflows so agents can operate with reliable business context, stronger governance, and less guesswork.

Role & Responsibilities

• Product Direction Assessment: Collaborated with the founding team to evaluate and shape the shift from the former Nexus developer platform into Fibinaci's enterprise-agent mapping direction
• Architectural Guidance: Delivered system-level recommendations on platform scope, technical feasibility, enterprise context mapping, and agentic workflows to establish the technical foundation
• Full-Stack Development: Performed hands-on frontend and backend development work, completing assigned tasks and implementing features while ensuring alignment with core business objectives
• Team Leadership: Led team members on technical implementations, provided guidance on development approaches, and coordinated work across multiple developers
• System Design Advisory: Provided high-level system design recommendations for cloud architecture, database optimization, and AWS infrastructure planning
• Strategic Planning: Advised on technical roadmap and implementation strategies to establish strategic clarity during initial engagement phase

Platform Direction & Technology

• Enterprise Context Map: Modeling company systems, data sources, permissions, workflows, and operational dependencies for agentic AI use cases
• Agentic Workflows: Designing workflows where agents can reason over business context and execute with clearer constraints
• Governance Layer: Emphasizing permissions, data boundaries, and operational accountability for enterprise AI adoption
• Full-Stack Platform: React, Node.js, TypeScript, cloud infrastructure, and API-driven architecture for product experimentation
• Legacy Nexus Concepts: Early developer-platform ideas informed the repositioning, but the current direction is centered on enterprise AI infrastructure

Team Structure & Management

• Architecture/Resource Team: Specializes in cloud databases, efficient resource management (storage and compute), AWS optimization, and client-side caching dynamics
• Algo/AI/Search Team: Handles social features, AI integration, algorithm needs, search functionality, and LLM implementation
• Full-Stack Developers: Each dev takes responsibility for feature sets, works on their branch, manages database integration, and submits pull requests for review

Weekly Development Workflow

• 1. Feedback Collection: Gather user feedback and prioritize feature ideas
• 2. Task Creation: Define and document tasks in GitHub Projects
• 3. Assignment: Distribute tasks to appropriate team members based on expertise
• 4. Development: Design UI (using shadcn and AI-assisted design), implement backend and frontend code, and test thoroughly
• 5. Review & Deployment: Submit pull requests, conduct code reviews, and merge approved changes

Impact & Contribution

Provided foundational technical guidance during the initial Nexus phase and helped shape its repositioning into Fibinaci. Contributed strategic and technical input around platform scope, agentic workflows, feasibility, system design, and the shift toward enterprise AI infrastructure that gives agents reliable organizational context.

Key Technologies

• Full-Stack: React, Node.js, JavaScript, TypeScript
• Cloud & Infrastructure: AWS, cloud databases, caching strategies, resource optimization
• AI Integration: LLMs, agentic workflows, enterprise context mapping, AI-powered development tools
• Collaboration Tools: Real-time collaboration, WebSockets, version control
• Management: GitHub Projects, agile workflow, code review processes, team coordination

AI Systems Engineer Intern | Talent Strategy Experts

Duration: Oct 2025 - Mar 2026

Location: San Jose, CA (Remote)

Platform: neuerraa.com

Mission & Vision

Working directly with CEO Zohra Veach to tackle the fundamental challenge of fixing the disconnect between talent strategy and business goals. The Neuerra Platform represents the future of talent strategy - an AI-powered system designed to eliminate inefficiencies in modern recruiting and HR processes through intelligent automation and data-driven insights.

Core Development Focus

• End-to-End AI Systems: Developing core AI systems that power automated candidate analysis and strategic interview coordination
• Multi-Tenant Architecture: Building scalable backend infrastructure designed to eliminate the disconnect between talent strategy and business goals
• Full-Stack Integration: Creating seamless data-rich experiences with React frontend modules and robust API integrations
• System Validation: Leading debugging and deployment efforts to ensure data isolation, reliability, and performance across organizations
• Platform Innovation: Contributing to the design and development of features that will define the future of talent strategy technology

Key Technologies

• Backend: Software architectural design, REST APIs, role-based access control (RBAC)
• Frontend: React.js, JavaScript for interactive dashboards and user flows
• Database: SQL for multi-tenant data management and optimization
• AI Integration: AI for recruiting, AI for HR automation, automated candidate analysis
• Architecture: Multi-tenant management, system validation, performance optimization

Software Development Intern | IDXExchange

Duration: Oct 2025 - Dec 2025

Location: Idaho (Remote)

Program Info: Fall 2025 Internship Program

Leadership & Project Scope

Leading a team of six interns in designing and enhancing web-based applications that power real estate property search and listing features. Overseeing full-stack development efforts integrating live MLS data through Cotality's Trestle API, guiding database design using SQL/MySQL for efficient data storage, retrieval, and normalization. Managing deployments and coordinating team collaboration for scalable real estate technology solutions.

Technical Leadership & Achievements

• Team Leadership: Leading a team of six interns, coordinating development efforts and ensuring project milestones
• MLS Integration: Overseeing connection to CRMLS (via Cotality's Trestle API) for real-time property data access and processing
• Database Architecture: Guiding SQL database design for optimal data storage, retrieval, and normalization of property listings
• Full-Stack Development: Managing both frontend interfaces and backend systems for comprehensive property search functionality
• Production Deployment: Overseeing deployment of fully functional applications to live server environment (InMotion Hosting with Linux, Apache/Nginx, PHP, and MySQL)
• Agile Management: Conducting weekly sprint reviews, presenting feature updates, and incorporating feedback from senior engineers and mentors
• Development Operations: Maintaining GitHub repositories, documentation, and reusable components to promote consistency, scalability, and effective team collaboration

Technical Stack

• Frontend: JavaScript/React, HTML/CSS, responsive design and modern frameworks
• Backend: PHP, Python for server-side logic and API communication
• Database: SQL/MySQL for data storage, retrieval, and normalization
• APIs: REST APIs (Trestle/Cotality), MLS data integration
• DevOps: Git version control, Linux (VPS/Server), Apache/Nginx web servers
• Hosting: InMotion Hosting with cPanel, production deployment and server management

AI Trainer - Software Developer | Outlier

Duration: April 2025 - Dec 2025

Location: Remote

Key Responsibilities

• Developed and evaluated prompts across technical domains, including code-based inputs in C++, Python, SQL, and JavaScript
• Reviewed AI-generated outputs for correctness, logic, and clarity in complex programming scenarios
• Produced improved reference implementations for various software development tasks
• Identified failure cases and explored prompt optimization strategies for enhanced AI performance
• Maintained 4.5/5 average task submission feedback across all assigned projects

ML Red Team Fellow | Handshake (MOVE Fellowship)

Duration: Aug 2025 - Apr 2026

Location: Remote

Key Responsibilities

• Collaborated with leading researchers to red team LLMs, exposing vulnerabilities, unsafe outputs, and injection risks
• Performed penetration testing with 200+ adversarial prompts, logging 50+ exploits and delivering mitigation strategies
• Conducted structured security assessments to identify model weaknesses and attack vectors
• Developed comprehensive reports documenting security vulnerabilities and recommended countermeasures

LLM: Evaluation Specialist | Alignerr

Duration: Sep 2025 - Apr 2026

Location: Chicago, IL (Remote) | Contract

Key Responsibilities

• Executed 4+ large-scale LLM evaluation and alignment initiatives spanning reasoning, coding, and function-calling domains
• Led rubric-based and human-preference assessments on thousands of samples to enhance model accuracy, coherence, and developer usability
• Contributing to key projects including Gravity Rubric Eval, Code Human Preference Eval, and Function Calling Evaluation
• Applied expertise in Large Language Model Operations (LLMOps) and Data Labeling to improve model performance and reliability

Skills & Technologies

CS Teaching Assistant | University of Illinois Chicago

Duration: Jan 2025 - May 2025, Aug 2025 - Dec 2025 (Spring & Fall Semesters)

Location: Chicago, IL

Key Responsibilities

• Selected for re-appointment due to exceptional performance and positive student feedback, demonstrating reliability and teaching effectiveness across Spring 2025 and Fall 2025 semesters.
• Assist in teaching CS 109: Programming for Engineers with MATLAB, supporting 250+ students per semester and 30+ students in weekly lab sessions.
• Grade and administer lab assignments with detailed feedback, proctor examinations, and conduct weekly office hours for student support.
• Develop course materials, facilitate tailored exam-week review sessions, and enhance student academic success through personalized guidance.

Engineering Success Program Mentor (ENGR 100/101) | University of Illinois Chicago

Duration: Jun 2025 - Present

Location: Chicago, IL (On-site)

Key Responsibilities

• Facilitate ENGR 100/101: Engineering Success Seminar series in sections of 24 students, collaborating with ECE mentor team across 10+ sections of freshman & transfer students.
• Create curriculum and projects, grade assignments, respond to course emails, and conduct weekly office hours during fall 25 and spring 26 semesters.
• Assist with Engineering Expo coordination and provide individualized academic support to enhance student engagement and success.
• Coordinate course delivery with fellow mentors while providing constructive, targeted feedback on projects and labs.

Code Coach | The Coder School

Duration: Jan 2025 - Feb 2026

Location: Lombard, IL

Key Responsibilities

• Mentoring students in Python, C++, Java, and web development technologies
• Designing custom lesson plans, projects, and coding challenges for varying skill levels
• Providing individualized feedback and support to foster student growth and interest in programming
• Teaching fundamental programming concepts and best practices

Student Projects Repository

Technical Lead | iCode School Franchise

Duration: Jan 2025 - Jun 2025

Location: Glen Ellyn, IL (On-site)

Key Responsibilities

• Served as lead STEM instructor for trial and group sessions for students ages 5-15
• Taught programming fundamentals, VEX robotics, digital logic, and technical problem solving
• Guided young learners through structured technical activities and hands-on STEM lessons
• Supported student engagement, classroom flow, and parent-facing communication during sessions

Scratch/Minecraft Instructor | Arrayscape Gaming, Inc.

Duration: Mar 2025 - Jul 2025

Location: Chicago, IL (On-site)

Key Responsibilities

• Introduced children to programming concepts through Scratch-based instruction
• Used Minecraft-centered learning environments to support engagement and computational thinking
• Delivered sessions across scheduled events throughout the Chicagoland area
• Helped students connect creative game-based activities with core programming logic

Deep Learning Research Intern | UIC ECE Department

Duration: Jun 2024 - Aug 2024

Location: Chicago, IL

Key Projects & Achievements

• Developed custom CNN architecture for wildfire detection achieving 20-30% accuracy improvement over baseline models
• Implemented complete video processing pipeline using PyTorch and OpenCV
• Collaborated with faculty and graduate students on computer vision research deliverables
• Conducted data preprocessing, model training, and performance evaluation

Additional Recognition

hls4ml Summer School | Northwestern University

Duration: Aug 2024

Location: Evanston, IL (On-site)

Program Overview

Selected as one of 20 students from UIC, UIUC, and Northwestern for a competitive three-day program hosted by CERN and the NSF focused on edge-AI hardware accelerators.

Technical Focus

• Gained hands-on training in hls4ml, converting PyTorch/Keras models into high-level synthesis workflows for FPGA deployment
• Applied FPGA accelerators through technical talks, tutorials, and a hackathon format
• Developed projects such as cartpole stabilization and a Pokemon card identifier
• Explored edge AI, IoT, embedded systems, and scientific computing applications

UI-UX/Machine Learning Intern | STEMAway

Duration: Jun 2023 - Sep 2023

Location: Remote

Key Projects & Achievements

• Participated in building and deploying ML pipelines for automated data collection and analysis using AWS cloud infrastructure
• Contributed to GPT API integration and web scraping technologies to streamline data workflows
• Collaborated on designing and launching the ByteMasters team website (https://bytemasters.softr.app/team-members) circa September 2023, showcasing team member profiles and project portfolios
• Assisted with UI/UX design improvements to enhance platform usability and accessibility for end users
• Gained hands-on experience with machine learning applications in web scraping and data processing workflows

Website Development

Private Tutor | Varsity Tutors & Grade Potential Tutoring

Duration: Aug 2025 - Present

Location: Remote & Greater Chicago Area

Varsity Tutors

• Provide personalized online tutoring for children and adults across 26+ approved subjects
• Specialize in AP Computer Science A & Principles, AP Calculus AB & BC, Programming Languages, and Artificial Intelligence
• Cover advanced mathematics including Calculus 2-3, College Algebra, Linear Algebra
• Teach Technology & Coding, Electrical and Computer Engineering concepts
• Strategic planning and educational development for diverse learning needs

Grade Potential Tutoring

• Conducted private in-person tutoring sessions for students of all ages in the Greater Chicago Area
• Supported test preparation, mathematics, classroom performance, and individualized academic development
• Built tailored learning strategies to improve confidence and academic performance

Private Tutor | Grade Potential Tutoring

Duration: Aug 2025 - Oct 2025

Location: Greater Chicago Area (On-site)

Role Overview

• Conduct private in-person tutoring sessions for students of all ages in Greater Chicago Area
• Focus on test preparation, mathematics, and individualized academic development
• Enhance classroom performance through tailored learning strategies
• Provide comprehensive support for academic growth and confidence building

Personal Projects Demo

Demo video coming soon! This will showcase interactive and creative web projects built with HTML, CSS, and JavaScript.

DSA_Library Demo

Demo video coming soon! This will highlight the features and performance of the C++ DSA library.

Embedded_Systems Demo

Demo video coming soon! This will showcase embedded systems projects, hardware integration, and real-time applications.

Fibinaci Agent Map

Status: Enterprise AI Infrastructure Concept

Former Website: nexusdevops.com

Role: Technical Consultant (Dec 2025 - Jan 2026)

About Fibinaci

Fibinaci is the renamed and refocused evolution of the former Nexus platform. The project is now positioned around enterprise AI infrastructure: mapping company systems, data, permissions, and workflows so agents can operate with reliable organizational context, better governance, and stronger operational accuracy.

Platform Vision & Mission

The original Nexus direction explored collaborative developer tooling. Fibinaci repositions that foundation around a sharper enterprise problem: giving AI agents a structured map of the business environments they operate inside, including systems, data boundaries, user permissions, workflow dependencies, and governance constraints.

Core Infrastructure Ideas

• Enterprise Context Map: Model company systems, data sources, permissions, workflows, and operational dependencies for agentic AI use cases
• Agentic Workflow Layer: Help agents reason over business context before taking actions across tools and systems
• Governance & Permissions: Represent access boundaries, approval paths, and safe execution constraints for enterprise environments
• System Integrations: Connect product, data, and workflow surfaces through API-driven architecture
• Legacy Nexus Foundation: Early developer-platform concepts informed the technical exploration before the business problem was narrowed

Technical Role & Contributions

• Product Direction Assessment: Collaborated with the founding team to evaluate and shape the shift from the former Nexus developer platform into Fibinaci's enterprise-agent mapping direction
• Architectural Guidance: Delivered system-level recommendations on platform scope, technical feasibility, enterprise context mapping, and agentic workflows to establish the technical foundation
• Full-Stack Development: Performed hands-on development work completing assigned tasks across frontend and backend technologies (React, Node.js, TypeScript), implementing features and ensuring code quality
• Team Leadership: Led team members on technical implementations, provided code guidance, reviewed work, and coordinated development efforts across multiple developers
• System Design Advisory: Provided high-level system design recommendations for cloud architecture, database optimization, and AWS infrastructure planning
• Strategic Planning: Advised on technical roadmap and implementation strategies to establish strategic clarity during initial engagement phase

Team Structure

Weekly Development Workflow

Enterprise Use Cases

Potential use cases include giving AI agents structured context for internal tools, customer workflows, data access rules, operational approvals, and cross-system task execution without relying on brittle assumptions about the organization.

Technical Stack

Impact

Provided foundational technical guidance during the initial Nexus phase and helped shape its repositioning into Fibinaci. Contributed strategic and technical input around platform scope, agentic workflows, feasibility, system design, and the shift toward enterprise AI infrastructure that gives agents reliable organizational context.

Neuerra Platform (Talent Strategy Experts)

Status: Enterprise Development (TSE Product)

Platform: neuerraa.com

Platform Vision & Impact

Contributing to the development of Neuerra (coming soon!), a revolutionary platform designed to eliminate the disconnect between talent strategy and business goals. Working directly with CEO Zohra Veach at Talent Strategy Experts, this flagship product represents the future of talent strategy - tackling fundamental challenges in modern recruiting through intelligent AI systems, seamless automation, and data-driven insights that transform how organizations approach talent acquisition.

Technical Contributions

• Multi-Tenant Architecture: Building and optimizing scalable backend infrastructure
• Role-Based Access Control: Implementing hierarchical permission systems
• Frontend Development: Enhancing React modules for improved user experience
• API Integration: Developing REST API endpoints and data flow optimization
• System Testing: Debugging and validation of multi-tenant workflows

Technical Stack

Professional Development

Full-Stack MLS Property Platform

Status: Live Production | SD6 Team @ IDXExchange (Fall 2025)

Live Demo: akbar.califorsale.org/search.php

Project Overview

Pro-bono real estate search platform engineered to integrate live MLS data from CRMLS (California Regional MLS) via the Trestle API. Led SD6 team of 6 developers in building a sophisticated platform with advanced property search, filtering, and visualization capabilities for 1,000+ California real estate listings.

Key Features

• Advanced Search & Filtering: Multi-criteria filtering (city, ZIP, price range, beds/baths, sqft) with real-time dynamic query building and sorting capabilities
• Dual View Modes: Grid and list layouts with interactive property modals featuring image galleries
• Session Management: Favorites system and recently viewed properties tracking
• Data Export: CSV export functionality for data analysis and real-time statistics dashboard
• Live Data Integration: Hourly automated synchronization with CRMLS via cron jobs
• Responsive Design: Mobile-first approach optimized for all devices with keyboard shortcuts

Technical Architecture

Database Schema

• Database: boxgra6_cali with primary table rets_property
• Core Fields: L_ListingID (PK), L_Address, L_City, L_Zip, L_SystemPrice, L_Keyword2 (beds), LM_Int2_3 (baths), LM_Dec_3 (sqft), L_Photos (JSON), L_UpdateDate
• Supporting Tables: token (OAuth storage), app_state (API offset/sync status)

Technology Stack

• Frontend: HTML5/CSS3 (modern gradient UI with flexbox/grid), Vanilla JavaScript ES6+
• Backend: PHP 7.2+ (server logic, API integration), Python 3.x (data processing)
• Database: MySQL 8.0+ with optimized queries and indexing
• DevOps: Linux VPS, Nginx web server, cPanel hosting management, automated cron tasks
• APIs: Trestle API (CoreLogic CRMLS), OAuth 2.0, OData Protocol

Team Leadership & Development

AArkboosted Website Auditor

Status: Live Production

Project Overview

A comprehensive automated website audit tool that performs security, SEO, and UI/UX analysis. Built with FastAPI and React 18, this platform automates manual auditing processes and generates professional PDF reports, reducing audit time by 40%+.

Core Features

• Comprehensive Analysis: Security vulnerabilities, SEO optimization, performance metrics, and UI/UX evaluation
• AI-Powered Scoring: Intelligent scoring system that provides actionable insights and recommendations
• Real Performance Data: Integration with Google PageSpeed API for accurate performance metrics
• Professional Reports: Automated PDF generation with detailed findings and improvement recommendations
• Security Scanning: Automated detection of common vulnerabilities and security misconfigurations
• SEO Analysis: Comprehensive SEO audit including meta tags, content optimization, and technical SEO
• Responsive Design: Modern, mobile-first interface built with Tailwind CSS

Technical Stack

Key API Endpoints

Installation & Usage

git clone https://github.com/ak23bar/AArkboosted-Website-Auditor
cd AArkboosted-Website-Auditor
pip install -r requirements.txt
uvicorn main:app --reload

cd frontend
npm install
npm run dev

Performance Impact

• Audit Speed: Reduces manual audit time by 40%+ through automation
• Accuracy: Consistent, repeatable results with comprehensive coverage
• Reporting: Professional PDF reports with actionable insights
• Scalability: Can handle multiple concurrent audits efficiently

Arkboostedads.com

Status: Live Production

Project Overview

Professional business website for Arkboosted, showcasing comprehensive digital marketing services and security-first web solutions. Features modern web architecture, conversion-optimized design, and strategic client engagement tools. Built with performance-focused technologies and hosted on Hostinger for enterprise-level reliability and global accessibility.

Core Features

• Service Portfolio: Comprehensive display of web audit and optimization services
• Client Testimonials: Social proof section showcasing customer satisfaction
• Contact Integration: Professional contact forms and consultation booking
• Responsive Design: Mobile-first approach ensuring optimal viewing on all devices
• SEO Optimization: Search engine optimized for maximum visibility
• Performance Metrics: Showcase of tool capabilities and success stories
• Professional Branding: Consistent brand identity and professional presentation

Technical Implementation

Business Integration

Development Approach

• Local Development: Built and tested locally for rapid iteration
• Version Control: Git-based workflow with planned GitHub repository
• Production Hosting: Deployed on Hostinger for reliability and performance
• Continuous Updates: Regular content and feature updates based on business needs

MATRIX - Voice AI Assistant

Status: Public Release

Project Overview

A sophisticated voice AI assistant that brings the Matrix aesthetic to conversational AI. Built with Python and powered by ElevenLabs' cutting-edge Conversational AI API, this application delivers real-time voice processing with an immersive cyberpunk interface.

Core Features

• Real-time Voice Processing: Continuous speech recognition with sub-second response times
• Neural Voice Synthesis: High-quality, natural-sounding AI voice responses
• Matrix-themed Interface: Immersive terminal experience with animated ASCII art
• Contextual Conversations: Maintains conversation history for coherent dialogue
• Cross-platform Audio: Optimized for Linux/WSL2 with PulseAudio integration

Technical Architecture

• Backend: Python application with Matrix-themed CLI interface
• AI Integration: ElevenLabs Conversational AI with WebSocket streaming
• Audio Pipeline: ALSA → PulseAudio → PyAudio → ElevenLabs
• Performance: Sub-200ms response times with optimized memory management
• Scalability: Modular architecture prepared for web application integration

Key Technologies

Core: Python 3.12+, ElevenLabs SDK, PyAudio, WebSocket Protocol
Audio: PulseAudio, ALSA utilities, Real-time audio streaming
Interface: ANSI color management, ASCII art system, Terminal enhancements
Infrastructure: Environment-based configuration, Comprehensive logging, Cross-platform compatibility

Pocket Brain - Offline Quantized LLM Assistant

Status: Active Development

GitHub: ak23bar/Pocket-Brain

Authors: Akbar Aman & Luke Abraham, 2026

Project Overview

Pocket Brain is a portable, fully offline conversational assistant for private brainstorming on constrained hardware. It targets CPU-only execution with a strict memory budget of ≤8GB RAM while maintaining predictable latency. The system integrates existing quantized open-weight LLMs through a local API, exposing structured interaction modes that improve output quality without increasing computational requirements.

Pocket Brain v1 is scoped as a personal, single-user tool. Design decisions prioritize determinism, portability, and explicit resource control over feature breadth. This is a systems engineering project focused on orchestrating quantized inference, enforcing bounded context growth, and providing structured prompting.

Core Constraints

• Offline-only: No outbound network calls required for inference or interaction
• CPU-only: Runs without GPU acceleration
• ≤8GB RAM: Total runtime memory must stay within this budget
• Portable local storage: Models and session artifacts reside on local disk/SSD

Architecture Components

• Clients: Web UI (HTML/JS) and CLI provide user interfaces
• API Gateway: Exposes /chat, /reset, /save endpoints
• Backend Orchestrator: Enforces modes, context caps, session state, and truncation policies
• Inference Engine: Quantized CPU-only model execution (GGUF format)
• Local Storage: Persists transcripts, session metadata, and model artifacts

Conversation Memory & Context Management

Large Language Models are stateless at inference time. Conversational "memory" is implemented by the application layer by resending a bounded subset of the transcript along with each new user message. Pocket Brain enforces a sliding window of the last N user-assistant exchanges (e.g., N=10), dropping older turns from the active inference context while persisting them to disk.

Interaction Modes

Pocket Brain enforces one mode per session, selected at session creation. Mode scoping stabilizes behavior on smaller models by avoiding multi-mode instruction stacking.

• perspectives: Generate alternative viewpoints and angles
• risks: Identify weaknesses, failure modes, and unknowns
• next_steps: Produce actionable next actions and priorities

Model Strategy (v1)

Pocket Brain does not train or fine-tune models. It integrates existing open-weight, instruction-tuned models that have been quantized (GGUF format) for efficient CPU inference.

• Qwen2.5 Instruct (3B): Default - balanced quality/speed, strong instruction following
• Phi-3 Mini (3.8B): Alt-fast - quick iteration, low memory pressure
• Mistral Instruct (7B): Optional - higher coherence, better complex reasoning

Implementation Milestones

• Milestone 1: End-to-end skeleton with API service, session creation, and transcript auto-save
• Milestone 2: Enforced constraints including fixed mode per session and sliding window history
• Milestone 3: Usability improvements with Web UI/CLI parity and "New Chat" rollover

Technical Stack

• Languages: Python (backend/orchestration), HTML/JavaScript (web UI)
• Models: Quantized GGUF models (Qwen2.5, Phi-3, Mistral)
• Inference: CPU-only quantized LLM execution
• Architecture: Local API service, session management, context truncation
• Storage: Local disk persistence for transcripts and metadata

What Pocket Brain is NOT

• Not a cloud service or multi-user platform
• Not a training or fine-tuning framework
• Not a production-ready consumer product
• Not a GPU-dependent system
• No model training, fine-tuning, or RAG in v1

Wildfire Detection Pipeline

Status: Research Project Completed

Project Overview

This repository supports video preprocessing, optical flow estimation, and inference using pretrained fire detection models (EfficientNetV2, ResNet18). Originally developed during a wildfire detection internship, this pipeline was used in conjunction with the Vision Mamba project for experimental evaluation.

Vision Mamba itself is not included, but this codebase contains adaptable tools for working with video datasets and classification models, and can be extended for use with other architectures or real-time fire detection tasks.

Technical Features

• Video Preprocessing: Prepare and segment video datasets for analysis
• Optical Flow Estimation: Motion analysis for fire detection in video streams
• Model Inference: Use pretrained EfficientNetV2 and ResNet18 models for fire classification
• Extensible Pipeline: Adaptable for other architectures and real-time detection tasks
• Integration: Originally used for experimental evaluation with Vision Mamba

Ultra Heat Sensor Alarm System

Status: Real-Time Embedded Safety Monitoring System

Platform: Texas Instruments Tiva C (TM4C123GH6PM) - ARM Cortex-M4F

Project Overview

Sophisticated embedded safety monitoring system providing real-time environmental monitoring through dual-mode operation: distance detection using ultrasonic ranging and temperature sensing with visual/audible alerts. Showcases advanced firmware engineering, hardware-software integration, and real-time event-driven architecture—all implemented in C and ARM assembly.

Key Highlights

• ⚡ Real-time multi-sensor processing with interrupt-driven architecture
• 🎯 Dual operational modes - distance detection and heat monitoring
• 🔊 Dynamic audio feedback with PWM-controlled buzzer system
• 📊 Multi-level alert system with color-coded LED indicators
• 🧮 Custom seven-segment display driver for real-time data visualization
• 🛠️ Mixed C/Assembly implementation demonstrating low-level hardware control

Distance Detection Mode (Ultrasonic Ranging)

• Precision Ranging: Up to 5+ feet with inch-level accuracy
• Safe Zone (≥5 ft): Green LED, no alarm
• Caution Zone (2-5 ft): Yellow LED, intermittent low-volume buzzer
• Danger Zone (<2 ft): Red LED, continuous high-volume alarm
• Real-time Display: Distance shown as XX:XX (feet:inches) format

Temperature Monitoring Mode

• High-Precision Sensing: 0.1°F resolution with continuous ADC sampling
• Digital Filtering: Signal conditioning for accurate readings
• Normal (<90°F): Green LED indicator
• Critical (≥90°F): Red LED + audible alarm
• Live Display: Temperature with decimal precision

Technical Implementation

Advanced Audio System

• PWM Tone Generation: Programmable pitch (C4-B4 musical scale)
• Volume Control: Three-level intensity (10%, 20%, 40% duty cycle)
• Pattern Differentiation: Separate alarm patterns for distance vs. temperature alerts
• Assembly Optimization: PWM configuration for precise timing

Event-Driven Architecture

• Non-Blocking Operation: Event scheduler and timer interrupts
• Concurrent Sensor Polling: No processor stalling
• State Machine: Efficient mode management implementation
• Power Optimization: WFI (Wait For Interrupt) instruction usage

Performance Metrics

Skills Demonstrated

• ✅ Bare-metal firmware development (no RTOS)
• ✅ Interrupt-driven event scheduling
• ✅ Hardware abstraction layer (HAL) design
• ✅ Mixed C and ARM assembly programming
• ✅ Direct peripheral register manipulation
• ✅ ADC sampling and signal conditioning
• ✅ PWM generation and control principles
• ✅ State machine implementation
• ✅ Sensor fusion (combining multiple inputs)
• ✅ Power-efficient design patterns

Embedded Systems Portfolio

Status: Coursework Collection

Project Overview

Comprehensive collection of ARM-based microcontroller projects demonstrating expertise in real-time systems, sensor integration, and bare-metal firmware development. Focus on low-level mixed C & assembly programming with direct hardware interfacing and peripheral control.

Note: While these projects were structured as group assignments, all code development, hardware integration, and system design were completed individually, representing solely my own work and technical contributions.

Core Technologies & Hardware

• Microcontroller Platform: Texas Instruments Tiva C Series (ARM Cortex-M4F)
• Development Tools: Code Composer Studio, TivaWare Peripheral Driver Library
• Sensor Integration: Ultrasonic sensors, temperature sensors, accelerometers, light sensors
• Communication Protocols: I2C, UART, SPI for device interfacing
• Output Devices: Seven-segment displays, LED arrays, LCD screens, PWM buzzers
• Power Management: WFI (Wait For Interrupt) optimization techniques

Programming Concepts & Implementation

• Interrupt-Driven Architecture: ISR implementation for timers, GPIO, ADC completion
• Real-Time Response: Event schedulers and state machines for concurrent operations
• ADC Sampling: Analog-to-digital conversion with signal filtering and calibration
• PWM Control: Pulse-width modulation for motor control, audio generation, LED dimming
• GPIO Management: Digital I/O control, debouncing, edge detection
• Timer Systems: Capture/compare modes, periodic interrupts, microsecond-precision timing
• Memory Management: Direct register manipulation, memory-mapped I/O, pointer optimization
• Mixed C/Assembly: Performance-critical sections in ARM assembly for efficiency

Key Skills Demonstrated

• Bare-Metal Development: Firmware without RTOS, direct hardware control
• Hardware Abstraction: Modular driver design for peripheral reusability
• System Architecture: Event-driven design patterns and state management
• Debugging Expertise: Logic analyzer usage, UART debugging, hardware troubleshooting
• Performance Optimization: Timing analysis, power consumption reduction, code efficiency

Project Highlights

Projects include sensor-based monitoring systems, real-time data acquisition, motor control applications, multi-device communication networks, and custom peripheral drivers. Each project emphasizes modular code structure, hardware-software co-design, and practical embedded systems engineering principles.

Web-Dev Collection

Status: Public Repository

Project Overview

A curated collection of mini website projects and creative ideas built with HTML, CSS, and JavaScript. Each project demonstrates modern design principles, interactive functionality, and responsive layouts.

Technical Implementation

• Frontend Technologies: Pure HTML5, CSS3, and vanilla JavaScript
• Design Focus: Modern UI/UX principles with responsive design patterns
• Interactive Elements: Dynamic content, animations, and user engagement features
• Cross-browser Compatibility: Optimized for all major browsers and devices

Featured Projects

• Bool Challenge Solutions: Creative website templates refined and enhanced for CodeHS competitions
• Interactive Demos: Animation showcases, layout experiments, and UI component libraries
• Responsive Layouts: Mobile-first design approaches with flexible grid systems
• Creative Solutions: Innovative approaches to common web development challenges

Key Features

• Modern Design: Clean, professional aesthetics with attention to detail
• Performance Optimized: Lightweight code with fast loading times
• Educational Value: Well-commented code for learning and reference
• Expandable Collection: Regularly updated with new projects and ideas

Enterprise RAG Support Agent

Status: Organization-Backed SaaS Support System

Project Overview

Scalable multi-tenant RAG system built for support automation across a large SaaS organization network. The system connected structured knowledge retrieval, LLM response generation, intake workflows, and escalation paths into a reusable support layer designed to reduce manual triage and improve answer consistency.

System Focus

• Retrieval Infrastructure: Vector-backed knowledge retrieval designed for fast, source-grounded support responses
• Multi-Tenant Support: Architecture shaped around a 700+ organization SaaS network with scalable query and intake workflows
• LLM Orchestration: Answer generation workflow with retrieval context, fallback behavior, and support escalation handling
• Operational Reliability: Emphasis on retrieval latency, query throughput, feedback handling, and unresolved-case handoff

Technical Stack

Impact

The project demonstrates applied AI infrastructure beyond a demo chatbot: retrieval performance, platform scale, operational handoff, and real support workflow design all mattered as part of the system.

hls4ml Edge-AI FPGA Projects

Status: Summer School Project Work / Edge-AI Hardware Acceleration

Project Overview

Hands-on FPGA and edge-AI project work completed through the hls4ml Summer School at Northwestern Engineering. The program focused on using hls4ml to convert machine learning models from frameworks such as PyTorch and Keras into high-level synthesis flows for FPGA deployment, with emphasis on latency, resource utilization, and hardware-aware model design.

Project Work

• Self-Balancing Cartpole Controller: Built and tested a controller that adjusted motor direction and speed on an FPGA-based physical testbed to keep a pole balanced while moving left and right.
• Pokemon Card Vision Reader: Developed and live-tested a computer vision model workflow for identifying Pokemon cards, applying edge-AI concepts to image classification.
• hls4ml Training: Learned the open-source hls4ml workflow for translating ML models into FPGA-oriented high-level synthesis implementations.

Engineering Focus

• FPGA-based machine learning inference and hardware acceleration
• Hardware-software co-design for real-time edge-AI systems
• Tradeoffs between model performance, latency, and resource utilization
• Control systems, computer vision, and embedded ML deployment workflows

References

CodeLingo

Status: Public Repository

Repository: ak23bar/CodeLingo

Project Overview

AI-powered code explanation tool built for students who can write code but struggle to read and reason through unfamiliar snippets. Paste code, select a language and explanation level, and receive a structured pedagogical breakdown designed for Python, JavaScript, and Java learners.

Core Features

• Level-Based Explanations: Beginner, student, and teacher explanation modes with adjusted vocabulary and depth
• Structured Output: Human translation, line-by-line explanation, execution walkthrough, purpose, common mistakes, and key concepts
• Streaming UI: Word-by-word response rendering for a more interactive AI product feel
• Quiz Mode: Generates comprehension questions with answer feedback and scoring
• Teacher Utility: Misconceptions-only toggle, Markdown export, and shareable snippet links

Technical Stack

• Streamlit frontend with dark theme and section-card styling
• Groq API with Llama 3.3 70B Versatile for structured reasoning and streaming
• Stateless Python app architecture using session state, environment variables, and Streamlit secrets
• Prompt reliability layer with heading validation, retry behavior, and safe fallback rendering

LLM Language Feedback API

Status: Public Repository / Applied AI API Challenge

Repository: ak23bar/intern-task-2026

Project Overview

FastAPI microservice that analyzes learner-written sentences and returns structured multilingual correction feedback. The system is designed around reliable LLM behavior: strict JSON output constraints, defensive parsing, schema validation, bounded repair retry, provider fallback, and cache-only-on-validated responses.

Core Capabilities

• Structured Feedback: Returns corrected sentence, correctness status, error objects, and CEFR-style difficulty in a predictable schema
• Multilingual Support: Handles learner feedback across Latin and non-Latin scripts with explanations in the user's native language
• LLM Reliability Layer: Uses defensive JSON parsing, enum checks, typed response construction, and one bounded repair retry for malformed output
• Provider Strategy: Routes through OpenAI first with Anthropic fallback for provider unavailability or call failure
• Operational Controls: Timeout-bounded provider calls, validated response caching, Docker support, and unit/schema/integration tests

Engineering Focus

Quantitative ML Research Platform

Status: Internship Research / Applied ML

Repositories: ml-track-ak23bar · spring-lab-ak23bar

Project Overview

Applied quantitative research and machine learning work completed through Tekly Studio's Crypt0nest lab. The work spans onboarding curriculum, ML4T research foundations, data pipeline auditing, strategy replication, technical indicators, and quantitative trading experiments.

Research & Engineering Focus

• Quant Strategy Work: Built and documented trading research exercises including Bollinger-RSI mean-reversion strategy work
• ML Foundations: Completed ML4T reading and replication exercises focused on market data, features, and model-oriented workflows
• Data Hygiene: Audited data architecture, survivorship risk, point-in-time accuracy, and pipeline reliability concerns
• Portfolio Metrics: Implemented indicators, portfolio metrics, verification notebooks, and experiment documentation
• Research Process: Maintained organized notebooks, proof artifacts, and commits across lab and ML track repositories

Technical Stack

• Python, Jupyter notebooks, pandas, NumPy, quantitative finance workflows
• Technical indicators, portfolio analytics, backtesting concepts, and strategy validation
• Data hygiene, feature engineering, market data structure, and reproducible research practices
• Git-based workflow across private Tekly Studio internship repositories

Computer Organization Portfolio

Tech Stack: Verilog, MIPS Assembly, Digital Logic, Computer Architecture

Repository: ak23bar/Computer-Organization

Project Overview

Coursework portfolio from ECE 366: Computer Organization, demonstrating digital logic design, hardware description languages, adder architecture analysis, and low-level MIPS assembly programming.

Digital Logic Components

• Full Adders: Structural and behavioral 1-bit full adder implementations
• Ripple-Carry Designs: 4-bit ripple-carry adder and subtractor with testbenches
• Carry Lookahead: 32-bit CLA using generate/propagate logic and hierarchical blocks
• Prefix Parallel Adder: 16-bit Brent-Kung style PPA for parallel carry computation

MIPS Assembly Components

• Iterative Fibonacci sequence generator with register-only computation and O(1) space
• Parity checker demonstrating low-level control flow and branch behavior
• Combined Fibonacci/parity program demonstrating function composition and register management
• Tradeoff analysis across instruction count, branching, memory access, and algorithmic efficiency

Engineering Concepts

• Hierarchical digital design from 1-bit components to wider arithmetic units
• Delay, area, and power tradeoffs across RCA, CLA, and PPA architectures
• Testbench design and verification methodology for hardware modules
• Processor-level reasoning through MIPS32 assembly and register discipline

AI Album Finder

Tech Stack: Python, Flask, REST API, Docker, JavaScript, Spotify API

Project Overview

AI-powered music discovery and analysis platform that leverages Spotify's Web API and artificial intelligence for deep music insights, mood analysis, and personalized recommendations.

Key Features

• Audio Analysis: Advanced audio feature extraction and mood profiling
• Smart Recommendations: AI-powered personalized music suggestions
• REST API: Flask-based backend with comprehensive endpoints
• Production Ready: Dockerized deployment with scalable infrastructure

Technical Implementation

• Integrated Spotify Web API for comprehensive music data access
• Implemented machine learning algorithms for mood and preference analysis
• Built responsive frontend with modern JavaScript and interactive UI
• Containerized application using Docker for consistent deployment

Student Projects

Tech Stack: Python, JavaScript, HTML/CSS, Web Development

Project Overview

A curated repository of educational projects designed and taught to students across multiple establishments, covering beginner to intermediate programming concepts.

Educational Focus

• Python Fundamentals: Core programming concepts and data structures
• Web Development: HTML, CSS, and JavaScript projects
• Interactive Projects: Games, calculators, and creative applications
• Collaborative Learning: Team-based coding exercises and challenges

Teaching Methodology

• Hands-on project-based learning with immediate practical application
• Progressive difficulty scaling from basic syntax to complex applications
• Regular code reviews and pair programming sessions
• Real-world problem solving and debugging techniques

Portfolio Website

Tech Stack: HTML, CSS, JavaScript, Responsive Design

Project Overview

Professional portfolio website demonstrating advanced front-end development capabilities through modern web technologies and performance optimization techniques.

Core Features

• Interactive Terminal: Fully functional CLI with 26+ Linux-style commands
• Dynamic Content: Typing animations with layout stability optimization
• Modal System: Comprehensive project and experience detail views
• Responsive Architecture: Mobile-first design with cross-device compatibility

Technical Architecture

• Vanilla JavaScript ES6+ with modular component structure
• CSS Grid and Flexbox for responsive layout implementation
• Semantic HTML5 with accessibility compliance (WCAG 2.1)
• Performance optimization including lazy loading and code splitting
• Cross-browser compatibility testing and progressive enhancement

Deployment & Performance

• Cloudflare Pages: Automated CI/CD pipeline with GitHub integration
• Global CDN: Edge caching and worldwide content delivery
• Performance: Optimized for fast loading and smooth user experience
• SEO Implementation: Semantic HTML and structured data
• Live URL: ak23bar.pages.dev

Data Structures & Algorithms Library

Tech Stack: Modern C++, STL

Project Overview

A comprehensive C++ library implementing fundamental data structures and algorithms, developed through coursework in Data Structures. Features practical applications from text processing to pathfinding, with emphasis on performance optimization and modern C++ practices.

Key Components

• Search Engine: Document search tool with AND/OR logic using inverted index, implemented with sets and maps
• Custom HashMap: Open addressing implementation with linear probing, dynamic resizing based on load factor
• BST Implementation: Complete binary search tree with recursive operations and visualization tools
• Graph Navigator: OpenStreetMap-based pathfinding using Dijkstra's and A* algorithms

Technical Highlights

• Implemented advanced data structures: linked lists, circular vectors, hash tables, and binary trees
• Focused on memory safety, OOP principles, and template design patterns
• Extensive performance analysis and benchmarking against STL containers
• Real-world applications including text analysis, path computation, and drawing systems

LeetCode Solutions

Tech Stack: Python, Java, Algorithms, Problem Solving

Project Overview

Comprehensive collection of LeetCode problem solutions designed for technical interview preparation and algorithmic skill development.

Key Features

• Multiple Languages: Solutions implemented in Python and Java with clean, readable code
• Detailed Explanations: Each solution includes approach explanation and complexity analysis
• Test Cases: Comprehensive test coverage to verify solution correctness
• Interview Focus: Optimized for technical interview scenarios and time constraints

Problem Categories

• Array and String manipulation with optimal time/space complexity
• Dynamic Programming solutions for complex optimization problems
• Tree and Graph algorithms including DFS, BFS, and shortest path
• Data structure design and implementation challenges