Samran E.

Senior AI Engineer with 7+ years building production AI in two areas: LLM systems (AI agents, RAG) and computer vision. I build end to end, from models and pipelines to Python/FastAPI backends and deploy on AWS, Azure, GCP and Runpod. Recent work • Production video segmentation & matting system (under NDA): 4K and 8K image segmentation and video matting, with high-accuracy person tracking across video using CLIP-ReID. The production solution outperformed SAM 3 and other open-source models on video segmentation and matting. • FundEvolve: a LangGraph multi-agent system that transcribes finance videos, fact-checks claims through RAG over Canadian regulations, and personalizes each verdict using the user's own financial data. • Reap Analytics: an AI agent chat interface for querying player statistics. AI agents & RAG ✓ Multi-agent systems with LangChain / LangGraph and MCP tools ✓ RAG chatbots and copilots on OpenAI, Claude and Mistral, with Pinecone, FAISS or Weaviate ✓ LLM fine-tuning (PEFT) Computer vision ✓ High-resolution (4K/8K) image segmentation and video matting ✓ Object detection, multi-person tracking and re-identification (CLIP-ReID) ✓ Video analytics and image processing with OpenCV ✓ Document AI: OCR and vision-language models for data extraction ✓ Training and deploying deep learning models in PyTorch Engineering ✓ FastAPI, Django · AWS, Azure, GCP, Runpod · DVC for model and data versioning Background: Senior ML Engineer at Cancer Clarity and Softech (computer vision specialist), ML researcher at FAST-NUCES, and founder of Rezunate AI. I work US Eastern hours and do my best work on ongoing engagements of 20–40 hrs/week. Send me your project and I'll reply with how I'd approach it.