Param Shah

solving AI Infra | ex-Microsoft AI Supercomputing | SR007

[Param Shah @shahparams]

@MLCatttt @t_ed_li @junyu_r @BanburyRoadAI @speedrun love it!

[Param Shah @shahparams]

@officialkivoart amazing stuff

[Param Shah @shahparams]

@gladkov_fillip @sigmagirlgrind @hf0 yo chilll

[Param Shah @shahparams]

@shivanijpatel DM me!

[Param Shah @shahparams]

@JayaGup10 DM me

[Param Shah @shahparams]

@nrol_ling so true!

[Param Shah @shahparams]

@nrol_ling

[Param Shah @shahparams]

@darrowoah Insane cost efficiency

[Param Shah @shahparams]

@17zhangw @OneWillAI @speedrun @EssenceVenture @lmwnshn @andy_pavlo @CMUDB LETS GOOO

[Will Zhang @17zhangw]

Excited to announce that @OneWillAI is backed by a16z @speedrun, @EssenceVenture, Tall VC, and amazing angels!

After a decade of working on machine learning for databases, @lmwnshn and I graduated with our PhDs this year from @andy_pavlo's @CMUDB. We got sufficiently annoyed with how AI agents worked that we decided to start a company to fix that.

OneWill’s core mission is to revive 50 years of database ideas for the AI era. This investment also allows us to give back a little to our roots in databases and open-source education. Stay tuned! We’re also open to more ideas, just join our Discord server (link below) or catch us at VLDB 2026 this week (Wed to Fri).

[Param Shah @shahparams]

@priya_kamdar @SutardjaCenter @Getthecheckpod @UCBerkeley @cjc @anikamirzaa @mayamshahh back 2 school

[Param Shah @shahparams]

Tomorrow we’re presenting at AI Faire, an invite-only a16z event at the brand-new Speedrun office in SF.

We’ve been heads down building in AI infra alongside an absurdly talented batch of founders.

Building in AI or thinking deeply about compute? Come say hi. Very few invites left. Reply or DM us.

[Param Shah @shahparams]

@karti_ai Can give it to you for even cheaper than that. We’re launching right now and would love for you to try it out!

[Param Shah @shahparams]

@pridiuksson @louszbd that’s a wild difference. what did GLM help unlock most on the H200 work? kernel changes, profiling/debugging, j faster iteration?

[Param Shah @shahparams]

@c_x_m_x @grok can help with that! just launched our neocloud and we provide H100/H200 compute at aggressive pricing, including short-term access. happy to get you started with 1 GPU and help with the setup. DM'ed you!

[Param Shah @shahparams]

@karti_ai can give you this for much cheaper (and even 2 IB connected nodes for sub $3/hr) as part of our launch! lmk if interested

[Param Shah @shahparams]

@net_termina still looking for compute? just launching our platform and we're providing 2 IB connected H200 nodes for way-below market pricing (as part of our launch)

[Param Shah @shahparams]

@xdotli bros a victim. i believe u dw

[Param Shah @shahparams]

@Getthecheckpod @ChristinaPhili5 @BainCapVC 💯

[Param Shah @shahparams]

@Getthecheckpod huge