Superlinked

This account is verified. Learn more

@superlinked

SIE serves small open models for everyday AI tasks. Find a model that meets your quality target, then call our API or run it in your own cloud.

superlinked.com Joined September 2019

311 Following 764 Followers 961 Posts

Latest posts 5

Pinned post

Superlinked @superlinked 5d

Help cut your AI costs with models built for the task!

Run 100+ task-specific models behind one production API.

Check out SIE Serverless in @svonava 's post below.

Daniel Svonava Verified account @svonava 5d

🚨 Huge news 🚨 today we are launching SIE Serverless - a managed inference product focused on small models. It’s useful for spiky workloads on 20+ task specific workloads across Search, OCR and Agentic execution.

Sign up for a $500 credits launch grant 👇

It’s powered by our open source inference engine that has received thousands of github stars in the last few months, battle-tested across finance, legal & tech. We believe that self-hosting is going to be critical given all the trust & security issues in AI, but practically there are always workloads where you’d like somebody else to deal with the GPU quota like synthetic data generation, evals, public research and beyond.

The team at Superlinked has been burning the midnight oil for months and I’m so proud of every single one of you - what a privilege to call you my team mates y’all!

We built it, now help us make some noise 🙌
A like, a comment or a repost would mean a lot! ❤️

View quoted post on X

Superlinked @superlinked 3d

Insanely good numbers...

Run the models and see the results for yourself: auth.superlinked.com/sign-up

Daniel Svonava Verified account @svonava 4d

"YOU NEED A BIG BOLD CLAIM IN YOUR LAUNCH"

So anyway, the evals we used for this and how robust are the findings, see: superlinked.com/blog/launch

View quoted post on X

Superlinked @superlinked 5d

A shelf camera can see the gap. Can your software prove which product is missing?

The proof is harder than the detection. Run OCR over the whole photo and you get every label on the shelf, and no idea which one sits under the hole. Spatial relationships are the real problem in shelf auditing.

This example solves them with four geometric rules: detect the gap, find the labels that overlap it vertically, pick the aligned pair, crop and upsample before OCR. On a real supermarket photo it identifies the missing product ("Panadol Child, 5-12Yrs Elixir 100ml") with zero hardcoded product knowledge.

Pure geometry, which is exactly what makes the audit reviewable by someone who wasn't in the store: superlinked.com/blog/retail-sh…

Superlinked @superlinked 6d

What's up NYC AI Founders!? If you're building own AI moat with your own fine-tuned models, here's a hackathon on Nov 7th you don't want to miss!

LOCALHOST NYC: Sovereign AI Hackathon

We want to bring together builders, researchers, and AI enthusiasts in NYC who are solving the hard challenges of owning the vertical AI stack including building proprietary task specific models in Legal, Finance, Healthcare, Cybersecurity, etc. and are fine-tuning and serving models they build!

Organized by Daniel Marzec, Brenner Spear and featuring @Superlinked, @cognee_ , Tinfoil, @darkbloomai , Variant, and more.

📍 The School, Brooklyn, NYC
💰 $10K in prizes + infrastructure credits
Let's build the future of open and sovereign AI together!
🔗 Register: luma.com/cognee-m0ow

luma.com LOCALHOST NYC: Sovereign AI Hackathon 🗽 · Luma No one should own your AI except for you. LOCALHOST is New York's Sovereign AI Hackathon, hosted at The School by dmarz and Brenner. Sponsored by Variant,…

Superlinked @superlinked Oct 27

AI that summarizes an accident report is easy. AI that never adds a claim the report doesn't make? That's the hard part, and it's the part that matters!

The test case: the NTSB's East Palestine report. Three wayside detectors recorded the same railcar bearing running hotter and hotter (38°F, then 103°F, then 253°F above ambient) in the hours before the derailment.

This pipeline reconstructs that sequence as a cited timeline: locations, times, temperatures, alerts, recipients. Every value traces to an exact span in the report, and the run halts if required evidence is missing.

The repo even kept its failed experiment (a schema-based approach that didn't survive contact with the material) committed as diagnostic evidence. That honesty is half the value.

For anyone building incident-review tooling: superlinked.com/blog/maintenan…