Researcher Profile
Editor reviewedHybrid Transformer–Mamba language models (Jamba)
ML researcher in the CTO office at AI21 Labs
A useful page because it points to the research-and-strategy side of AI21 rather than only the product or engineering side, especially where model evaluation and new architectural bets get shaped at the CTO-office level.
Organizations
Labs
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
The ideas, systems, and research directions that make this person worth knowing.
01
ML research in AI21 Labs CTO office
02
Research strategy around hybrid model systems
03
Connecting exploratory work to product-facing model releases
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
Canonical papers, project pages, or repositories that anchor this profile.
Additional papers, projects, or repositories that help flesh out the profile.
Additional links that help verify and flesh out this profile.
People worth exploring next because they share topics, labs, or source material with this profile.
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.
One of the higher-signal people to know in the hybrid-LLM line because he sits at the point where AI21’s research architecture, long-context systems work, and real product deployment meet.
A distinctive page in this AI21 cluster because she brings a linguistics and human-evaluation angle to model work, especially around user interaction, multilingual language behavior, and how LLM performance gets tested in practice.
Worth knowing because his work links earlier dense-retrieval research to later MRKL and Jamba systems, which makes his page a good bridge between classic NLP retrieval and newer hybrid LLM stacks.
One of the better pages in this cluster because it connects AI21 alignment work to concrete retrieval and grounding research rather than leaving "alignment" as a vague label.
An especially valuable page for understanding how AI systems get judged in practice, because it puts human evaluation and rubric design at the center rather than treating them as an afterthought to model building.