Mayank J.

I'm a Forward Deployed AI Engineer. I take AI ideas into your environment and turn them into systems your team actually uses: LLM assistants, RAG pipelines, AI agents and computer vision, from first conversation to running in production. With 8+ years in ML and software engineering, I work as an embedded part of your team. I learn your data, workflows and constraints, build quickly, and stay until the system is reliable. WHAT I BUILD - LLM applications and chatbots: customer support assistants, internal knowledge bots and RAG over your documents, using OpenAI, Anthropic, Llama, Mistral and LangChain - AI agents and automation: multi-step workflows that connect LLMs to your APIs, databases and tools, with guardrails - Computer vision: real-time video analytics on CCTV/IP camera streams, including shoplifting and theft detection, object detection and tracking (OpenCV, YOLO, PyTorch) - Forecasting and predictive analytics: demand forecasting and time-series models tied to business decisions - Production deployment: Docker, CI/CD and MLOps, plus evaluation, monitoring and handover to your team RECENT WORK - Retail theft detection system: real-time computer vision that analyses store camera feeds, detects suspicious behaviour such as concealment and shoplifting, and alerts staff instantly - AI customer support chatbot: LLM assistant that answers routine customer queries from the company knowledge base and hands complex cases to human agents - AI demand forecasting platform: predictive models that forecast product demand to support inventory and planning decisions - Financial GPT: LLM assistant for financial data analysis and Q&A - Object detection and tracking system, delivered on time (5-star client review) HOW I WORK 1. Discovery: understand the problem, data and success metric 2. Prototype in days, not weeks, so you see it working early 3. Harden for production: evaluations, error handling, deployment in your stack 4. Hand over with documentation so your team owns it STACK: Python, PyTorch, TensorFlow, Hugging Face, LangChain, OpenAI API, RAG and vector databases, OpenCV, YOLO, Docker, MLOps Have an AI idea that needs to work in the real world, not just in a demo? Message me with what you're trying to achieve and I'll reply with how I'd approach it.