Shaheryar B.

I build AI agents that work in production, not just in demos, and I train teams to do the same. By day I'm a Manager, Data & AI at Capgemini, where I have designed, built and deployed 20+ production AI Solutions for the Department of Health – Abu Dhabi: grounded data assistants, bilingual document generation, rubric-based assessment and recommendation engines, and an AI-assisted fraud-surveillance product. Before that I put ML and GenAI into production in maritime SaaS. WHAT I CAN BUILD FOR YOU • AI agents and assistants with tools, memory and approval steps (Claude, Azure OpenAI, Microsoft Agent Framework) • RAG / document Q&A over your PDFs, policies or knowledge base, with evaluation so you know it's right • Data-grounded assistants: plain-English questions over your database or Databricks • Document generation: AI drafts, humans approve, output in your Word templates • Production hardening of an existing prototype: tests, guardrails, logging, Docker, CI/CD HOW I WORK • Grounding over guessing: answers come from your data, not the model's imagination • The model extracts, code decides: anything with numbers, scores or rules is computed and repeatable • A human approves anything consequential • An evaluation set from day one, so changes don't silently break things • Clear handover: docs, tests, and a walkthrough call TRAINING AND WORKSHOPS (live online) • Building Production AI Agents: hands-on, for engineering teams • AI at Work: Copilot, agents and Claude for non-technical staff • AI Adoption & Governance for leaders Sessions run in UK/EU afternoons or US mornings, or on weekends. CREDENTIALS · Claude Certified Architect – Professional (Anthropic) · Microsoft Certified: Agentic AI Business Solutions Architect · Microsoft Certified: AI Transformation Leader · Azure AI Engineer Associate · Databricks AI Engineer Associate · MSc Data Science (First Class). STACK · Python, FastAPI, React/TypeScript · Claude, Azure OpenAI, Azure AI Foundry, Microsoft Agent Framework, Copilot Studio · RAG, vector databases (Weaviate), LangChain · Databricks, MLflow · Docker, Kubernetes, GitHub Actions, Azure DevOps. Send me a short description of your use case and the data it touches. I'll reply with how I'd approach it, what could go wrong, and a fixed-price option.