Muhammad Sameer K.

I build AI systems that do real work inside a business: agents that act across your tools, knowledge and document systems that run on your actual data, automations that take repetitive handling off your team, and AI features built into your product. Most projects I take on start where the demo ends. Messy documents, permissions, integrations with systems you already use, cost, and the question every buyer should ask: what happens when the model gets it wrong? 50+ AI projects delivered. I stay your technical lead from first call to deployment, and I bring in a small engineering team when a build needs more hands. WHAT I BUILD AI agents and agentic workflows Agents that read, decide and take action across CRMs, ERPs, email, databases and internal APIs. Built with LangGraph and tool calling, with the parts that matter in production done properly: permissions, human approval steps, retries, logging and evaluation. Used for operations, support, research, sales ops and back-office work. Enterprise RAG and document intelligence Knowledge systems over PDFs, Word files, spreadsheets and scanned documents. OCR, extraction, classification, chunking, hybrid search, reranking, and permission-aware retrieval. Answers you can trace back to the source document, and structured data you can push into the next step of a workflow. AI workflow automation For document-heavy, repetitive processes: intake, routing, extraction, validation, approvals, notifications and audit trails. n8n, Zapier, MCP or custom Python, depending on what the process needs. The goal is simple. Your people handle the exceptions, not every single item. AI product and backend engineering Taking an AI capability and turning it into software. FastAPI backends, multi-tenant architecture, auth, usage and cost tracking, model routing across OpenAI, Anthropic and open-source models, Docker and AWS deployment. For SaaS teams adding AI to an existing product, and founders who need the AI core built right. SOME OF THE WORK - Enterprise AI collaboration platform: lead AI engineer on a secure workspace giving teams access to 250+ AI models, shared workspaces and RAG over private company knowledge. - AI financial advisory platform (UK-regulated): backend and AI engineering for personalised investment insights and portfolio recommendations. - Agentic ERP/MRP system: LangGraph architecture with 25+ agents covering orders, engineering, CRM, QA and inventory. - Document intelligence pipelines: PDF, DOCX and XLSX ingestion with OCR, embeddings, vector search and retrieval analytics. Some of this was built for clients or inside larger teams. Happy to walk you through exactly what I owned on each. HOW I WORK - Workflow first. I want to understand your current process, where the time goes and where human review matters before recommending agents, RAG or automation. - Built to run, not to demo. Evaluation, error handling, monitoring and cost control are part of the scope. - You talk to me directly. Regular demos, clear written updates, no hand-offs to someone you've never met. - Fixed-price milestones when scope is clear, hourly for discovery and ongoing work. STACK Python, FastAPI, LangGraph, LangChain, LlamaIndex, OpenAI, Anthropic, Ollama, PostgreSQL, Redis, Chroma, Pinecone, FAISS, Docker, AWS, n8n, Zapier, MCP. If you have a process that eats hours of manual work, a knowledge base nobody can search properly, or an AI feature you need built into your product, send me a few lines about how it works today. I'll tell you straight whether AI fits, and what I'd build first.