My Projects
A collection of my work in AI/GPU computing, full-stack development, and algorithms.
From high-performance systems to useful, production-ready products.
Favorite Projects
A few projects I'm particularly proud of
ModelPulse
Self-improving GPU-accelerated RAG system - track LLM research with hybrid retrieval & local inference
An open-source, fully local retrieval-augmented AI system that keeps you up-to-date with the fast-moving world of LLMs. Automatically collects updates from trusted sources (arXiv, Hugging Face, OpenAI, Anthropic), builds semantic search indexes, and generates grounded summaries with citations. Features hybrid retrieval (BM25 + FAISS vector search), cross-encoder reranking, and GPU-accelerated local inference with Qwen/Mistral/Llama models. Includes RAGAS evaluation metrics for faithfulness, latency tracking, and an adaptive self-improving feedback loop. Everything runs locally with no API calls required (optional OpenAI key for advanced metrics).
Key Features:
- •Hybrid retrieval: BM25 + FAISS vector search with cross-encoder reranking
- •100% local GPU inference - no API calls needed
- •Self-improving with RAGAS evaluation & adaptive tuning
- •Tracks arXiv, Hugging Face, OpenAI, and Anthropic updates
- •Grounded Q&A with source citations
- •Docker-based deployment with NVIDIA GPU acceleration
Push Swap Tester
Testing suite for School 42's push_swap - validates sorting algorithms with performance benchmarks & debug visualization
An open-source testing utility for the School 42 push_swap project. Provides comprehensive validation through basic parsing tests, identity checks, small stack operations (3-5 elements), and configurable performance testing for medium-to-large stacks. Features multiple debugging modes including manual input visualization, random number generation, and permutation sequences. Generates detailed test reports with clear status indicators (OK/KO/TLE/ERROR), stores test traces, and includes memory leak detection capabilities. Built to help students debug and optimize their sorting algorithms efficiently.
Key Features:
- •Comprehensive test suite: parsing, identity, small stacks (3-5)
- •Configurable performance testing with iteration counts
- •Multiple debug modes: manual input, random generation, permutations
- •Clear result reporting: OK/KO/TLE/ERROR status
- •Memory leak detection support
- •Open-source tool for School 42 community