Umair A.

I build machine learning systems that handle work usually done by hand - reading documents, detecting objects in images, extracting and organizing data - and get them running reliably in production. I've worked in software for 6 years and now focus on applied ML and AI. WHAT I WORK ON: ➝ RAG & LLM automation — answering questions over your own documents, classifying and summarizing text, and pipelines that cut repetitive manual work ➝ Computer Vision — object & brand detection (YOLOv8–11), image segmentation, OCR for scanned documents and photos ➝ NLP — metadata extraction (SigLIP), multilingual NER, sentiment analysis ➝ Python backends — FastAPI/Flask APIs, ETL pipelines, SQL/NoSQL optimization Because I came from mobile development, I can also take a model the rest of the way, into an API, an automation, or a Flutter app - not just leave it as a script. STACK: Python, PyTorch, YOLO/Ultralytics, OpenCV, Hugging Face Transformers, LangChain, OpenAI & open-source LLMs, pgvector/FAISS, pandas, FastAPI, PostgreSQL, MongoDB, Docker, Flutter, Firebase, Git. If it helps, we can start with a short call or a small paid task so you can see how I work.