Rachel StClair

Founder & CEO of @servamind · PhD in AI · building the data layer AI runs on: encode once: train, fine-tune & serve for less. Papers + open source this fall.

[Rachel StClair @RachelStClair7]

I've spent my career on the underlying math of AI — and the question of what it actually takes to get to a Data-from-TNG level of intelligence.

Now I'm building @servamind: a model-agnostic data layer for AI. Encode your data once, and the same file works across training, fine-tuning, and inference — less runtime, less power, targeting ~1/10th the cost.

This fall: two research papers and an open-source release, with a repro kit so you don't have to take our word for anything.

Follow along — I'll be posting about the tokenizer tax, why data preparation is AI's most ignored bottleneck, and what we find as we run the benchmarks.

[Rachel StClair @RachelStClair7]

@_ScottCondron Chimera soon. Would love to hear your needs. We’ll build for you.

[Rachel StClair @RachelStClair7]

@Sahil100X @jitendarma73662 @Adarsh_VMore @ashutosh_sao @deepanshuliv @Deepjyoti100x @kirat_tw @100xDevs @100xSchool @rohanBuilds @AnkurJangid02 Would love to chat. DM me please!

[Rachel StClair @RachelStClair7]

@baseten Would auto backrop optimization help you?

[Rachel StClair @RachelStClair7]

@adisingh I would use https://t.co/w6kqp1gBSI

[Rachel StClair @RachelStClair7]

@racerfunction Whats the secret?

[Rachel StClair @RachelStClair7]

@sridharfyi https://t.co/t7vLm4lNmC

[Rachel StClair @RachelStClair7]

@asmah2107 Hi! We would love to connect

[Rachel StClair @RachelStClair7]

@raymondmfeng Would love to share notes!

[Rachel StClair @RachelStClair7]

@ypatil125 super cool

[Rachel StClair @RachelStClair7]

@lukaskuhn77 like if you have .mp4, .wav, .avi, can you feed all 3? or you still have to preprocess

[Rachel StClair @RachelStClair7]

@divaagurlxw > yes

[Rachel StClair @RachelStClair7]

@llama_index if only there was a universal encoder that turned any format into a shared AI-ready format

[Rachel StClair @RachelStClair7]

@k44yej @Microsoft @MicrosoftAI Would love to see how you manage data pipe for your multimodal

[Rachel StClair @RachelStClair7]

@BredisGeorge We’re building a better tokenizer and I’d love to see how it works for you

[Rachel StClair @RachelStClair7]

@Keanaalabre Added you in linkedin

[Rachel StClair @RachelStClair7]

@AstraiaAI Yes, several approaches been around for a while just needs to be scales up

[Rachel StClair @RachelStClair7]

@ben_sage Building because we thought people would need it, just checked in to make sure, seems they still do.

[Rachel StClair @RachelStClair7]

@TheGeorgePu Hey we have the same question! We’re cooking a massively faster 70b for several families. So would love to resurface this or see what ya’ll really want.