I work on AI systems end to end, from the underlying mathematics, algorithms, and computer vision pipelines that process raw data, to training, fine-tuning, quantizing, and optimizing models for efficient GPU and CPU inference. On the agentic side, I design the architecture that brings these components together through context engineering, RAG, tool use, memory, and state management. I also build the backend infrastructure that powers these systems, including asynchronous and parallel processing, distributed workloads, and scalable service architectures. Beyond individual systems, I build AI agent platforms on top of ERP systems, connecting AI capabilities with real-world business workflows. My focus is on engineering AI systems that are efficient, reliable, production-ready, and scalable.
- Languages & frameworks:
Python,FastAPI,asyncio,Celery,Redis,Git - LLMs & orchestration:
LangGraph,LangChain,MCP,vLLM - Observability & evaluation:
LangSmith,Langfuse,Arize,Phoenix,OpenLit - CV & ML:
OpenCV,NumPy,PyTorch,OpenVINO,YOLO,Model Quantization - Inference & deployment:
GPU/CPU Inference,Docker - Datastores & infra:
PostgreSQL,MariaDB,ChromaDB,Qdrant - ERP & integrations:
Frappe/ERPNext