Practice
Software, including AI systems.
The work is ordinary in the best sense: understand the problem, build the system, put it in production, keep it honest. AI is one of the tools we use, when it earns its place.
-
Custom software
Products, internal tools, and the unglamorous systems that actually run a company. Web services, APIs, data pipelines, and the operational layer around them.
-
Applied AI
Language models, retrieval, agents, and evaluation wired into real workflows. We care about failure modes, data boundaries, and whether the thing is usable on a Tuesday afternoon.
-
Delivery and counsel
Architecture, implementation, and technical judgment.