
editor @ command line // comms director @microsoft
San Francisco Bay Area
Joined January 2009
- Tweets2.2K
- Following181
- Followers213
- Likes198
msft audited DeepSWE and Terminal-Bench 4.0, flagged 22 flawed tasks, and worked collaboratively to help improve the benchmarks. commandline.microsoft.com/it…
11
what @Microsoft learned rebuilding @novonordisk’s clinical trials data agent on Foundry: a simpler operating model, a stronger session isolation, and a direct path from tracing to evaluation and optimization // cc: @AIFoundryDevs commandline.microsoft.com/fr…
47
the thinkingbox team looks at expanded model evaluations, the price of dependable runs, and reinforcement learning on terminal-state rewards commandline.microsoft.com/th…
24
from empty repo to global GA in < 5 months, plus a runtime migration from python to rust commandline.microsoft.com/ho…
30
🚀🚀🚀
Today at #GitHubUniverse we laid out how we're evolving GitHub for the agentic era. Check it out: gh.io/gh-universe
1
60
a look at the evolving SDLC from @timbozarth 🤖 commandline.microsoft.com/ne…
14
msft’s FoSSE team is building and running AI agents that transform some of the world’s largest codebases, including Windows and Azure commandline.microsoft.com/ag…
35
decision model meets microsoft autopilot (née miscrosoft scout) commandline.microsoft.com/bu…
55
AI can code, but humans need to define success, interpret evidence, & turn judgment into changes the team can test. spend time on journey evaluation as you shorten the distance from idea to product ship. commandline.microsoft.com/mi…
16
msft & @CarnegieMellon are betting that the career of software development is here to stay—and that the junior devs we train today will be the senior engineers of tomorrow commandline.microsoft.com/pr…
32
agentic AI for network incident response, with defined identity, authorization, tooling, and human-approval boundaries commandline.microsoft.com/ai…
1
63
coming soon, GitHub Copilot will determine when a task is best handled by on-device AI and when it should leverage cloud-scale models. commandline.microsoft.com/lo…
21
decouple storage from compute → scale each independently, recover faster from failures, & match capacity to demand
Really proud of the work that the team has been doing. We’re rearchitecting our core Git infra to stay ahead of growth and support the future of software development.
gh.io/building-git-infrastru…
34
Beyond violation rates: Are your AI evaluations measuring the right things? commandline.microsoft.com/st…
32
Building an agentic system and a learning loop around customer IP commandline.microsoft.com/ka…
34
a little trip down memory lane on command line today, looking at a decade of @Azure content understanding + document intelligence commandline.microsoft.com/az…
1
61
Shared design components aren’t enough to ensure a seamless product experience. CoreAI’s designers are creating consistency by embracing simplicity over complexity commandline.microsoft.com/de…
29
agents are moving into production faster than the governance around them. enter: agent hooks! commandline.microsoft.com/ag…
35
How do you translate a high-level safety, policy, or product requirement into evaluations and controls that reliably govern agent behavior in production? 🤔 commandline.microsoft.com/sa…
26

