Selected work

Shipped, not slideware.

A few representative engagements, anonymized. The pattern is the point: taken from prototype to production and handed to the team to run.

A retail-technology team needed recommendations and AI agents that could carry real traffic, not just a demo.

BuiltRecommender systems and AI agents across the data and serving path.

ResultDelivered as production systems, not prototypes.

recommenders · AI agents · production serving

An enterprise telecom needed its machine-learning work to run reliably and repeatably in production.

BuiltML workflows on Azure Machine Learning and Databricks.

ResultStandardized and shipped into their production environment.

Azure ML · Databricks · MLOps

An overwhelmed manual customer-support queue.

BuiltA retrieval-augmented (RAG) support assistant, taken from prototype to production.

ResultDeployed to handle first-line support and handed to the team to run.

RAG · LLM · MLOps · handover

You work directly with Isaac as principal on every engagement. When scope needs it, Nazmi scales with a small network of vetted senior collaborators — never a junior handoff.

Start

Bring Nazmi in on the work that has to ship.

Tell Nazmi what exists today, what is not working, and what needs to happen next.

Start a conversation or book a 20-minute call →