Talks
I speak about what actually breaks when agents meet production traffic: cost that runs away quietly, failures that will not reproduce, and workflows that have to recover on their own. Available for conferences, podcasts, and internal engineering forums.
Upcoming
-
Debugging the Agent That Refuses to Fail the Same Way Twice
How to capture every non-deterministic input to an agent loop in an append-only execution log, replay the failed run step by step, and turn that replay into a reproducible eval. The talk works through a real incident rather than a toy example.
Conference talks
-
FinOps for AI Agents: Who Spent All the Tokens?
A practical look at cost management and token optimisation for non-deterministic agents: where the spend actually goes, how to make it visible mid-run, and how to design agents that stay economical as they scale. The talk introduced TokenOps.
-
Your Agent Failed in Prod. Good Luck Reproducing It.
Why non-determinism makes agent failures so hard to chase, and what record-and-replay has to capture before a production incident can be reproduced on a laptop.
-
Spec-Driven Development with Spec Kit
Treating requirements as the first architectural control surface, and where that structure earns its keep versus where it turns into tax.
Microsoft AI Driven Engineering series
Internal sessions I presented in Microsoft's AIDE series, covering agents across the development lifecycle. Not publicly recorded.
-
Azure DevOps SWE Agent
Architecture, custom agents, MCP server integration, and prompting practices that survive contact with a real repository.
-
Azure SRE Agent
Onboarding, incident automation, ICM and DevOps integration, and sub-agent builder workflows.
-
Spec Kit, SWE Agent and Playwright Testing Agents
Brownfield migration with Spec Kit, automated pull request creation, and Playwright Planner, Generator and Healer agents that write and self-heal test cases.
-
The Agency Platform and AI-First Repositories
Microsoft's first-party agent platform, and the structure, docs, and context engineering that make a repository genuinely agent-ready.
-
GitHub Copilot CLI and SDK with Liquid MCP
Copilot SDK and CLI, MCP server integration, workflow handling, and compliance guidance.