Now
Oct 2026At Datadog, I am a Staff engineer in Datadog AI Research (DAIR), working on the Trained Agents for Observability team. I build simulation environments, RL training loops, and evaluation infrastructure used to post-train agents for SRE incident response. Before DAIR, I led the query and connector platform those agents use for data access.
Outside work, I play chess and Age of Empires II, build and maintain a few personal tools, and help with open source when I can. I'm a Substrait committer and occasionally contribute to Apache Beam, Calcite, DataFusion, and Trino.
Projects
More on GitHubA few tools I build and use for AI-assisted engineering and research.
Agent session portability
OmniSession
Search, resume, fork, and move local coding-agent sessions across nine harnesses while preserving history and verifying each transfer.
Web-to-LLM extraction
Copy as Markdown
A browser extension and userscript that turn pages, docs, dashboards, and chats into structured Markdown for LLM context while preserving links, tables, code, and metadata.
More projects
Theme-aware visual primitives and render pipeline for coding agents. Write TSX scenes, then render diagrams, dashboards, presentation frames, and GIFs in dark and light themes.
Fast native Mermaid rendering in Go for SVG and PNG output, without a browser process.
A curated collection of browser userscripts that I created and maintain.
Current setup
Oct 2026- Operating systems
macOS (company laptop) 
Linux Mint (personal desktop)- Coding harness
Codex CLI / GPT 6.1 Sol 
Claude Code / Opus 5.5- Clipboard manager
CopyQ- Text editor
Sublime Text- Text expansion
Espanso
cmux 
Limux