AI Engineering
Designing and shipping production LLM systems — from prompt architecture to evaluation harnesses and cost-aware serving.
Outcome · Reliable AI features that survive contact with real users.
I help enterprises turn sprawling, messy data estates into intelligent, AI-powered platforms that scale — pairing 15+ years of data engineering depth with a builder's obsession for autonomous, agentic systems.
About
For more than fifteen years I've worked at the seam where raw data becomes business advantage — designing the platforms, pipelines, and architectures that let global organisations trust their numbers and act on them in real time. My path has taken me from Mumbai and Pune to Manchester, London, and now Dublin, and across insurance, financial services, and enterprise technology.
In recent years my focus has shifted to the next frontier: autonomous, agentic AI. I design systems where LLMs, RAG pipelines, and coordinated AI agents don't just answer questions — they do the work: generating documentation, enforcing data quality, and running pipelines end to end. Frameworks like LangGraph, CrewAI, and MCP are my current toolkit for turning that vision into production reality.
Read my full storyExpertise
Deep specialisation across the modern data and AI stack — from enterprise architecture to autonomous agents.
Designing and shipping production LLM systems — from prompt architecture to evaluation harnesses and cost-aware serving.
Outcome · Reliable AI features that survive contact with real users.
Building autonomous agents that plan, use tools, and coordinate to complete multi-step engineering work without a human in the loop.
Outcome · Systems that do the work, not just describe it.
Metadata-driven pipelines and frameworks that ingest, transform, and serve data reliably at enterprise scale.
Outcome · Pipelines that stay correct as volume 10×'s.
Lakehouse, medallion, and event-driven architectures designed for governance, observability, and long-term scale.
Outcome · Foundations teams can build on for a decade.
Performance-tuned, cost-optimised Snowflake platforms with secure data sharing and warehouse governance.
Outcome · Faster queries at a fraction of the spend.
Cloud-native data platforms on AWS and Azure — infrastructure as code, serverless, and secure by default.
Outcome · Elastic platforms that scale to demand.
Selected Work
A selection of platforms and systems I've architected and delivered.
A unified, governed data foundation for a global enterprise.
An autonomous agent that detects, explains, and remediates data issues.
Config-over-code ingestion that scales to hundreds of sources.
A grounded assistant over the organisation's internal knowledge.
Where AI agents build and operate the data platform itself.
Testimonials
“One of the sharpest data architects I've worked with. Zubair sees the whole system — he'll fix the immediate problem while quietly re-designing the thing so it never happens again.”
Whether you're modernising a data platform or bringing agentic AI into production, I'd love to hear what you're working on.