🛰️ I'm currently working on

Building production-grade Agentic AI systems using LangGraph and RAG architectures — currently focused on reducing LLM hallucination through evaluation-driven retrieval optimization (RAGAS, cross-encoder reranking).
👥 I'm looking to collaborate on

Open-source LLM tooling, agentic AI frameworks, and MLOps/LLMOps pipelines — especially projects involving multi-agent orchestration, RAG evaluation frameworks, or fine-tuning workflows.
🤝 I'm looking for help with

Scaling self-hosted LLM inference (vLLM/Ollama) for cost-efficient production deployment, and deepening my understanding of distributed training at scale.
🌱 I'm currently learning

Advanced agentic architecture patterns (multi-agent orchestration, GraphRAG), LLM quantization techniques (AWQ/GPTQ), and RLHF/DPO fine-tuning methods.
💬 Ask me about

LLM fine-tuning (LoRA/QLoRA), RAG pipeline design, Agentic AI with LangGraph/LangChain, MLOps/CI-CD for ML systems, or building production AI from scratch.
⚡ Fun fact

I built a transformer-based LLM completely from scratch in PyTorch — zero pre-built libraries — just to understand exactly what happens inside every layer.