I'm a Senior Staff AI/ML Engineer and Architect focused on the messy, high-stakes work of bringing agentic AI into production. I help lead the AI/ML architecture and development for a portfolio of agentic systems deployed across unclassified and classified programs.
What I've learned building these systems is that the model is rarely the problem. Agent quality lives in the layer underneath. It's about how context is assembled and pruned each turn, how tool calls are orchestrated and recovered from, how the action surface is scoped so an agent can be useful without being dangerous, and how observability, guardrails, and identity are designed in from day one rather than retrofitted under deployment pressure. A lot of my time goes into that layer, including the design of MCP servers and model-legible tools that hold up under real workloads.
That work rests on more than two decades in software and data engineering, increasingly centered on machine learning over the last decade-plus. Long before agents, I was profiling training loops, building data platforms, and researching, designing, and testing model architectures, from CNNs to RNNs and transformers. The architectural decisions behind a reliable agent platform are downstream of knowing how data flows, how models train, and how systems fail.
The other half of the job is people. I lead a cross-functional team of AI, software, machine learning, and data engineers, and I take mentoring seriously: pairing on hard problems, helping junior engineers grow into senior ones, and establishing the practices that let a team move fast without breaking what matters. I also serve as a senior lead on our AI Enablement initiative, driving adoption of agentic tooling and engineering practices across the organization.
If you're building something hard with agents, I'd be love to compare notes.