Get ideas into production
When an idea has earned it, you take it all the way. If nothing's been built yet, you design and build it with the person who came up with it. If there's a prototype or a pilot, you work out what it does well and where it falls down. Then you harden it for real use: security, data handling, testing, monitoring, cost control, and a named owner. If only the original author understands it, it doesn't go live.
Own the architecture
You own the technical design of everything the team builds and keep it hanging together, so each new piece adds to a shared set of components instead of sitting on its own. The question you keep asking is whether Pfizer's platform already does this, or whether it's really specific to OCEO.
Write and review code
You'll write and review code most days. AI tools do most of the implementation. You keep the architecture and the reviews in your own hands, and knowing where that line sits is a big part of the job.
Refine the agentic software factory
The team builds with an agentic software factory. Every piece of work goes through the same steps: specify, plan, build, review, ship. AI agents do most of the building, and engineers check what comes out. You'll keep improving it: tightening the specification and review steps, choosing the tools and agents it runs on, and fixing whatever keeps sending work back for another pass.
Work with the platform teams
You're the technical contact for Pfizer's enterprise AI platform teams. You pick up their new capabilities as they land and tell them what OCEO needs. When a platform slips, you change the plan rather than building a shadow version of it.
Get things approved
Solutions go through the Architecture Review Board, the Security Review and the AI Review Board, and you help with the privacy and risk assessments. Nothing touches real data or real users until it's approved. Wherever AI output acts in a system or leaves OCEO, a person signs it off. Access stays as narrow as the work allows.
Take over partner work
Sometimes external partners or colleagues build something and hand it over. You agree up front what a good handover looks like, check it before you accept it, and fix the gaps. OCEO shouldn't end up supporting something it can't change.
Run it in production
Once it's live, you run it: monitoring, evaluation results, cost, incidents, rollback. You should know what each solution costs and what it gives back, and step in early if quality starts to slip or the bill starts to climb.
Build with the people who had the idea
Work with the colleagues who bring ideas in, so they know what they're taking on when it goes live. Make it easy for them to extend things themselves, with the same controls applied whoever builds the addition.