
- Joined
- Jan 2025
- How they joined
- Directly
- Before Cognition
- Sequoia CapitalSummer Analyst
- Status
- Former
“internally we gave devin a credit card and people would have devin buy stuff for the office. you'd often see packages addressed to "Devin {last_name_of_designated_human}"”
- Public headline says "Cognition, Sequoia, Stanford".
- Personal website: https://andrew.md/; this comes from the profile summary and preserves the person's own positioning.
- Previous roles include Human Software Engineer at Cognition; Venture Partner at Contrary; Summer Analyst at Sequoia Capital; Software Engineer at Etched; Software Engineering Intern at Pika; Software Engineer at Exa; Software Engineering Intern at LangChain; ML Research Intern (comp bio focus) at San Diego Supercomputer Center.
- Education history includes Bachelors, Computer Science, Stanford University; Canyon Crest Academy; Master of Science, Computer Science, Stanford University.
- Location is Stanford, California, United States.
First intern, engineer
Contract
Contract
Building infrastructure for finetuning of large language models
Did not complete
Public school in San Diego
. @midjourney Scanner interactive 3d model I made! fun piece of trivia: I interned for @JMRothberg who created the ultrasound-on-chip tech behind the scanner great work @DavidSHolz + team! play with the model below (thx Devin for carrying) https://t.co/9JfWgAleza
@babarganesh I'd argue that AI hit during the freshman year for the class of 2026 - ChatGPT came out in November 30, 2022 and by the sophomore year, was good enough to do a lot of coding. Cursor launched sometime in 2023 and Devin in 2024. It's now 2026.
@cognition https://t.co/JxW7HwX0Xl
Mythos 5 / Fable 5 blows away the competition on @cognition FrontierCode Diamond (newest and least saturated SWE benchmark) FrontierCode tasks 👇 https://t.co/N3WSD2ea3D
can't find a screenshot but this vaguely reminds me of when @Adhyyan added a feature to Devin where if you were angry at it, it would give you a small refund
every developer I know uses both Claude Code and Codex, switching between them when either quota runs out Devin Desktop is the easiest way to do this
I made this when I was an intern lol
with the spacex @cursor_ai news, the ai coding agent space just gets more exciting genuinely curious if cursor will still have access to anthropic models (anyone rmb what happened to og windsurf when the openai acquisition was brewing) seems like some doors are opening and some
lowkey would "Spacebar" be a fun name for SpaceXaiCursor? Cursor and Spacebar are both things you use to interact with your computer spacebar also makes sense bc cursor 3 creates a space to organize agent sessions (windsurf 2.0 as well!) or maybe elon wants to save Spacebar fo
ever since the windsurf acquisition in july, people have been asking when this would happen! it's finally here https://t.co/heL6YG4uBC
a couple years ago, we were just barely able to eke out useful coding ability from models - so every time the latest model came out (gpt-3, gpt-3.5, 4o), it just made sense that you'd immediately switch all your coding tasks over to the latest and greatest. now, the models seem
i also often use devin for random non-coding tasks internally we gave devin a credit card and people would have devin buy stuff for the office. you'd often see packages addressed to "Devin {last_name_of_designated_human}"