Ali Hamza Azam — Computer Scientist and ML Engineer
Introduction
Skills and Projects
Chattr
Real-time secure chat with end-to-end RSA public-key encryption, delivery/read receipts, and PostgreSQL row-level security. Built on Nuxt 3, Vue 3, Socket.io and Supabase. Built with Nuxt 3, Vue 3, Socket.io, Supabase, TypeScript.
Zeta
A readable programming language with English-like syntax — full interpreter pipeline (lexer, recursive-descent parser, semantic analyzer, tree-walking interpreter) in Java. Runs entirely in the browser via a CheerpJ/WASM playground, no backend. Built with Java, CheerpJ, TypeScript, Compilers, Automata.
IPFS DHT Simulator
Chord-style distributed hash table — O(log N) finger-table routing, B-tree file indexing, and node join/leave with key redistribution. C++ core with an interactive browser visualizer of the ring. Built with C++, TypeScript, DHT, Chord, B-Tree.
Pulselog
Uptime-monitoring service with scheduled health checks, automatic incident management, notifications, and public status pages. Built with Laravel 12, Blade, and PostgreSQL. Built with Laravel 12, PHP, Blade, PostgreSQL.
Linkvault
Personal bookmarking service with automatic metadata extraction, collections, tagging, and full-text search. Built with Laravel 12, Blade, and PostgreSQL. Built with Laravel 12, PHP, Blade, PostgreSQL.
json2relcsv
A C tool that parses arbitrary JSON and flattens it into a set of related CSV tables (one per object shape, linked by foreign keys), driven by a hand-written Flex/Bison grammar. Compiled to WebAssembly for an in-browser playground. Built with C, Flex/Bison, WebAssembly, Emscripten.
Montage AI
Turns a single text prompt into a fully narrated, multi-scene visual-novel video. A LangGraph multi-agent pipeline drives Gemini, Google Cloud TTS and FFmpeg, with a FastAPI + React UI and a natural-language edit agent. Built with Python, LangGraph, Gemini, FastAPI, React, FFmpeg.
CRAG Bench
Benchmarks six advanced RAG strategies — RAG Fusion, HyDE, CRAG, Graph RAG, Multi-Query and RRR — against the noisy CRAG web corpus, with a shared embedding index and per-pipeline evaluation. Built with Python, RAG, LLMs, FAISS, Streamlit.
Multimodal RAG
Multimodal Retrieval-Augmented Generation system combining CLIP embeddings, FAISS vector search, and an MLX-powered Mistral LLM over mixed text + image corpora. Built with Python, CLIP, FAISS, MLX, Mistral, RAG.
StudyBuddy
Research comparing Baseline LLM, RAG, and Fine-tuned approaches for building AI tutors that respond appropriately to frustrated, curious, confident, and tired students. Built with Python, LLMs, RAG, PyTorch, MLX.
Parallel SSSP
Efficient parallel implementation of Single Source Shortest Path algorithms in C++, with OpenMP and MPI support for distributed and scalable graph processing. Built with C++, OpenMP, MPI, HPC, Graph Algorithms.
repomon
Mission control for orchestrating parallel AI coding agents — Claude Code, Codex, and Aider — across many repos, branches, and git worktrees. A Rust/Ratatui TUI backed by a tmux daemon. Built with Rust, Ratatui, tmux, TUI, Git.
Pacman Assembly
Complete Pac-Man game built entirely in Assembly language, demonstrating deep understanding of computer architecture, memory management, and low-level programming. Built with Assembly, x86, Low-Level Programming.
Diffusion Transformer SiT
PyTorch implementation of SiT-B (Scalable Interpolant Transformer) for diffusion-based image generation, with REG modules and a v-prediction objective. Built with Python, PyTorch, Diffusion, Transformers.
QuizVerse
Full-stack web application for creating, sharing, and participating in quizzes. Features robust authentication (local and Google OAuth), customizable user profiles, and an admin dashboard. Built with Vue.js, Node.js, Express, MongoDB, OAuth.
Waste Detection DL
YOLOv8 object detection and U-Net segmentation for waste classification on the TACO dataset, applying deep learning to environmental sustainability. Built with Python, YOLOv8, U-Net, PyTorch, Computer Vision.
ABOUT
System specifications and capabilities
Computer scientist building systems that scale — across distributed computing, machine-learning infrastructure, and developer tooling.
Focused on making complex systems legible through visualization and clean interfaces. I write code others can read, and I'm happiest somewhere between a compiler, a GPU, and a research paper.
CONTACT
Initialize communication protocol
Interested in collaboration or have a question? Feel free to reach out through any of the following channels.