Log inSign up
Image
user avatar
Diptanu Choudhury
@diptanu
Founder @tensorlake, building sandbox infra for agents. Past - Built Nomad @hashicorp, AI Infra @meta, Container Scheduler tech lead @linkedin and @netlfix
San Francisco, CA
tensorlake.ai
Joined December 2007
881
Following
4,758
Followers
RepliesArticlesMedia
  • user avatar
    Diptanu Choudhury
    @diptanu
    4h
    I forgot we are on @computesdk's storage benchmark now! Storage is our new product, we have been dog fooding it ourselves by mounting cloud volumes backed by this backend in our CI workflows. It shares build caches across jobs to speed up builds and tests.
    Image
  • user avatar
    Diptanu Choudhury
    @diptanu
    Aug 2
    Kind of a bummer that the recent paper on Europa is predicting there is very little chance to reach the sub surface ocean easily from the surface. Wonder what the europa clipper mission find when it gets there in 2030.
  • user avatar
    Diptanu Choudhury
    @diptanu
    Aug 2
    The Xerox PARC of AI?
    user avatar
    Tibo
    @thsottiaux
    Aug 1
    Replying to @_chenglou
    I was part of that team. Basically ChatGPT one year before it came out. Called LMChat and then another codename. Google was too nervous to release it and DeepMind was blocked from shipping products that could disrupt Google. I think about this a lot.
  • user avatar
    Diptanu Choudhury
    @diptanu
    Aug 2
    Most bare metal server companies seem to provide AMD CPUs while hyperscalers are on Intel. Why is that? It's a problem for us because we can't restore memory snapshots of sandboxes created on AWS on AMD servers.
  • user avatar
    Diptanu Choudhury
    @diptanu
    Aug 2
    We are seeing once a user builds their CI infrastructure on Tensorlake, usage keeps going up and up as they start using or building software factories. Claude and Codex are more of a limiter at this point to acceleration because they still need humans in the loop for the most
    user avatar
    Diptanu Choudhury
    @diptanu
    Aug 1
    CI is one of the bottlenecks for builders shipping 100x more code today. You can now build scalable GitHub Runner infrastructure programatically on Tensorlake sandboxes, scaling out to thousands of sandboxes when PRs are updated to run tests, then scaling back down to zero!
    Image