4 sources
A sensible page to keep because his name appears directly on the original Jamba paper, giving users another concrete entry point into the people who built AI21’s hybrid architecture.
Researcher Profile
Editor reviewedErez Safahi
Hybrid Transformer–Mamba language models (Jamba)
Researcher behind AI21's Jamba hybrid-model work
Useful because it turns one of the anonymous-looking Jamba authors into an actual person page, which makes the hybrid-model line easier to understand than treating it as a single monolithic team output.
Organizations
AI21 Labs
Labs
Topics
About This Page
This profile is meant to help you get oriented quickly: why this researcher matters, what to read first, and where to explore next.
Last reviewed
March 18, 2026
Known For
The ideas, systems, and research directions that make this person worth knowing.
01
Work on the original Jamba release
02
Hybrid Transformer-Mamba model development
03
Public open-weight model releases at AI21 Labs
04
Hybrid Transformer–Mamba language models (Jamba)
05
Jamba: A Hybrid Transformer-Mamba Language Model
06
Jamba-1.5: Hybrid Transformer-Mamba Models at Scale
Start Here
Canonical papers, project pages, or repositories that anchor this profile.
Related Researchers
People worth exploring next because they share topics, labs, or source material with this profile.
4 sources
A valuable page in this cluster because his public role description is unusually specific: post-training, steerability, and AI-generated evaluation data are exactly the kinds of practical problems strong researcher pages should make discoverable.
Start HereAlan Arazi
5 sources
A strong long-tail researcher page because his public profile explicitly points to factual knowledge and grounding, which are much more useful signals than another generic AI21/Jamba placeholder.
Start HereDor Muhlgay
4 sources
A worthwhile head-page upgrade because it gives one of the quieter Jamba contributors a concrete place in the stack: the pre- and post-training work that turns a hybrid architecture into an actual usable model.
4 sources
Worth surfacing because he sits inside the original Jamba author group, which helps make the AI21 hybrid-model story legible at the contributor level instead of only at the company level.
3 sources
A useful long-tail page because it exposes another named contributor to AI21’s hybrid architecture work rather than leaving the profile buried inside a shared cohort summary.