Sabri Eyuboglu (@EyubogluSabri)

Pinned post

Most times I use an agent it spends minutes rereading all my stuff – spending dollars and energy just to gather the context it needs to understand my work.

But it won’t always be this way. We can train that understanding into the model – amortizing across tasks.

We started Engram to build the systems required for a future where there are billions of models, each learning from users and deeply understanding their work.

I’m so grateful and excited to be working on this problem with this incredible team

Engram @EngramLab Jun 23

x.com/i/article/2069…
View quoted post on X

Sabri Eyuboglu @EyubogluSabri Nov 4

A student takes notes for an open book exam and shares them with a friend. They both use the notes during the exam, yet only the one who made it aces

Similarly, for language models, neural and text memory stack

Engram @EngramLab Nov 4

In our new blog, we share some new science investigating how RL complements text-based memory by letting models learn how to use their notes.

It turns out that training can be cheap compared to building notes, and stays surprisingly robust even through millions of tokens of edits.

View quoted post on X

Sabri Eyuboglu @EyubogluSabri Nov 4

love this from @dylan__lim @Drewwad @reyna_abhyankar and team – despite all the talk of self-improvement platforms and splashy results, relatively little open work on the technical tradeoffs you confront in building one

Simran Arora Verified account @simran_s_arora Nov 4

Introducing the Together Compiler: continual compilation for AI inference! I’m super proud to finally share this work from our research team! ✨

Today’s inference is “Polymorphic” or shape-shifting, it doesn’t sit still. On OpenRouter, 80% of traffic goes to models released in the last three months and the top ten most popular open models each use a different architecture. Our customers constantly change their usage behaviors. Together Compiler leverages agents to continually optimize our deployments immediately as these changes occur.

Together Compiler research highlights include:
1. 252x better throughput: TKCC lets agents try 252x more optimization experiments per hour compared to prior agent inference optimization systems.
2. State-of-the-art TPS across InferenceX workloads we tried: TKCC outperforms the best InferenceX engine across all tested settings in just 24 hours & 2 nodes per compile. Including 1111 TPS on MiniMax M3.1 B200 and 1205 TPS on DeepSeek v4.1 Flash GB300 at batch size 1. NVIDIA, AMD, Trainium2.
3. Continuous recompilation: TKCC currently serves many trillions of production tokens per day and adapts to traffic shifts within just a few hours of recompilation.
4. Bitter lessons: TKCC’s design builds on 2+ years of building frameworks to accelerate AI Inference, starting from ThunderKittens and Megakernels. Compilers have always used abstractions and IRs, but we show that agentic compilers now reach higher performance ceilings without abstractions.

From an incredible team @reyna_abhyankar @dylan__lim @realDanFu @togethercompute!!

View quoted post on X

Sabri Eyuboglu @EyubogluSabri Oct 29

Been a while since I saw a result as cool as in-context tokenization

Albert Gu Verified account @_albertgu Oct 28

I've always believed that intelligence should come from simple, general building blocks that learn their own abstractions directly from raw data; in other words, the bitter lesson. I think we're finally seeing a real step towards AI that can learn from all data.

Language modeling on raw bytes is a microcosm of this ideal, and to me it's a central problem in AI: if we can't build LLMs that learn end-to-end from bytes, how will we handle the rest of the world's data that we understand even less?

Last year we introduced hierarchical networks (H-Nets) as a first step. Our new version finds up to 3-5x efficiency gains in controlled scaling law experiments, and at larger scale matches an open-source model trained on 11T tokens with ~10x less data and compute. The posts below walk through what we found and why we think it works.

View quoted post on X

Sabri Eyuboglu @EyubogluSabri Aug 17

We had Qwen 3.8 study all of the docs in Harvey’s synthetic law firm - the more it studies, the fewer tokens it needs at inference time to respond to the queries

Engram @EngramLab Aug 17

Today we're publishing our first research blog, Understanding a Law Firm through Study.

We're sharing a glimpse of a future where agents are trained with native memory:

View quoted post on X

Sabri Eyuboglu

Latest posts 5