I'm Mohammed Aintomar, an AI Engineer focused on RAG systems, LLM applications, and backend engineering.
I enjoy turning AI concepts into practical systems — from document ingestion and embeddings to hybrid retrieval, reranking, citations, evaluation, and production-oriented APIs.
Currently focused on:
- 🧠 Retrieval-Augmented Generation (RAG)
- 🤖 LLM-powered applications
- 🔎 Hybrid search and reranking
- 📊 AI evaluation and retrieval metrics
- 🐍 Python backend engineering
- 🗄️ PostgreSQL + pgvector
- 🐳 Docker and Linux
- 🌐 REST APIs and integrations
AI-powered codebase intelligence system
CodeAtlas is a portfolio-focused AI system for exploring and asking questions about source-code repositories using retrieval and LLMs.
- Repository ingestion from ZIP files and GitHub repositories
- AST-aware source-code chunking
- Sentence-transformer embeddings
- PostgreSQL + pgvector
- HNSW vector search
- Hybrid semantic + keyword retrieval
- Reranking
- Source citations with file paths and line references
- Retrieval evaluation queries
- Flask API
- Docker Compose
Embedded security contribution
Contributed to EmbeddedOS eBoot with shared image slot boundary validation.
The contribution strengthens validation around image headers, image size, and TLV boundaries across secure boot and recovery-related paths.
| Area | Focus |
|---|---|
| RAG | Retrieval pipelines, grounding, citations |
| Search | Dense retrieval, BM25, hybrid search, RRF |
| Reranking | Candidate retrieval → reranking → top-k context |
| Embeddings | Semantic representation and vector search |
| Vector DB | PostgreSQL + pgvector + HNSW |
| LLMs | Prompting and structured generation |
| Evaluation | Recall@K, groundedness, retrieval quality |
| Backend | Python, Flask, REST APIs, PostgreSQL |
| Infrastructure | Linux, Docker, Git |
I'm interested in contributing to projects around:
- AI engineering
- RAG and developer tools
- Backend systems
- Open-source infrastructure
- Embedded systems and security
Building practical AI systems, one repository at a time.