I build production AI that does real work.
I work inside a business, on the systems it already has, from the first conversation to handover.
About
I'm Gbolagade, an AI engineer based in London. I build the software a business runs on day to day: internal tools, customer-facing products, and the AI systems behind them.
Most of my work is agentic. I have built an assistant that triages email, runs a calendar and places calls, and a hub that joins a company's internal apps into one assistant with a memory layer behind it. Some of the work is plain engineering, such as a back-office integration or an internal tool, because that is often what the problem needs.
I have done this in-house and on contract for clients in pharmacy, wholesale and retail. With clients I work as a forward deployed engineer. I sit down with the founder or the operations manager, build on the systems they already have, run the launch, and train their staff before I hand it over.
Everything I ship has evals, cost tracking and an audit trail, so the people running it can see what it did and what it cost. In my own time I run open-weight models on a laptop to see how far a small one goes.
What I Build & The Impact
Business software & internal tools
The day-to-day systems a business runs on: internal tools, dashboards, integrations, and automations, in custom code or on Power Platform, that remove manual work and connect the systems you already use.
Production LLM & agent systems
RAG pipelines, multi-agent orchestration, agent memory, MCP servers, and AI co-pilots that take real actions behind confirmation. Evals and structured-output validation keep them predictable in production.
Forward deployed delivery
With clients I work inside the business. I scope the problem with the founder or operations manager, build on the systems already in place, run the launch, then train the staff and hand over. I have done this across customer operations, ecommerce catalogues, content, and back-office workflows.
Observability, governance & open source
Latency, cost, and output-validity monitoring with GDPR-aligned redaction, plus open-source tools and local-model experiments that I publish. You can see what a system did, what it cost, and who approved it.
Projects
Connect
Building something, or thinking about it? Book a 1:1 and let's dig in.
Gbolagade Ishola