Antar Basil

I’m an AI engineer focused on turning LLM concepts into production systems—especially RAG and agentic workflows. I build end-to-end pipelines that cover document ingestion, chunking, embedding, retrieval, and generation.

In freelance work, I designed hybrid retrieval (semantic + keyword) to improve accuracy, then deployed systems with FastAPI + PostgreSQL (pgvector) + Docker. I’ve also developed vector search capabilities using PostgreSQL (pgvector) and ChromaDB.

I architected agentic AI workflows with autonomous task orchestration using LangGraph patterns, including multi-step reasoning with tool calling. I also designed a serverless-first architecture with async processing to support scalable AI services.

For model performance, I benchmarked Whisper model variants and achieved 40% latency reduction through model selection and optimization, including -70% latency via kernel-level tuning. Alongside my AI/ML work, I bring software engineering and reliability experience from managing infrastructure for 50+ devices with 99.9% uptime.