Prime Intellect careers: remote jobs, culture and where they hire

Prime Intellect has raised more than $150M across three rounds. A $5.5M seed in April 2024 was co-led by CoinFund and Distributed Global with Compound, Collab+Currency, Protocol Labs and angels including Clem Delangue of Hugging Face and Dylan Patel of SemiAnalysis. A $15M round in February 2025 was led by Founders Fund with Menlo Ventures and angels including Andrej Karpathy, Tri Dao and Emad Mostaque, taking the total past $20M. The $130M Series A in July 2026 was led by Radical Ventures with NVIDIA Ventures, Intel Capital, Dell Technologies Capital and Iconiq, at a $1B valuation according to TechCrunch, with more than 18 angels including John Schulman, Aaron Levie, Matthew Prince and Harrison Chase.

Prime Intellect builds infrastructure for training, evaluating and deploying AI models, and uses that same stack to train its own open models. The product, Lab, bundles GPU compute, RL environments, secure sandboxes, hosted evaluations against 100+ open-source models and deployment into one post-training system, so a company can turn its own workflows and feedback data into a model it owns. Underneath sits a compute layer that aggregates H100, H200, B200 and B300 capacity from many providers into one console, with SLURM or Kubernetes orchestration and a spot market for reselling reserved clusters. The company also runs the Environments Hub, a community library of more than 2,500 RL environments built on its open-source Verifiers library.

The company started in January 2024 as a decentralized training project and shipped its milestones in public: INTELLECT-1, a 10B model trained over 42 days on up to 112 H100s across five countries and three continents (November 2024); INTELLECT-2, a 32B model trained with globally distributed RL (May 2025); and INTELLECT-3, a 106B mixture-of-experts model trained for two months on 512 H200s, with weights, the PRIME-RL trainer, the environments and the technical report all released (November 2025). By the July 2026 Series A it reported more than 6,000 customers and over $100M in annualized revenue, with Ramp, Zapier, Character.AI, Browserbase, Goodfire and Arcee among the named users; Ramp trained a 35B model on Lab that it says beat frontier models on spreadsheet search at a fraction of the cost.

Vincent Weisser, CEO and co-founder. German entrepreneur from the decentralized science world: co-founded the longevity research DAO VitaDAO in 2021 and was Chief of Ecosystem and AI at Molecule, the biopharma IP platform, from 2021 to 2023, after co-founding the exchange dex.blue. Studied software engineering, product and design at CODE University of Applied Sciences in Berlin, which he left before graduating.

Johannes Hagemann, CTO and co-founder. Research engineer at Aleph Alpha from early 2022 to 2023, where he helped build the German lab's distributed training framework from scratch. He says he became "scaling-pilled" after reading the Scaling Hypothesis in 2020 and decided with Weisser at the end of 2023 to go all in on Prime Intellect. M.Sc. in IT-Systems Engineering from the Hasso Plattner Institute in Potsdam and a B.Sc. in Computer Science from TU Dortmund.

Prime Intellect is one of the few places where a small team trains open frontier models and sells the tooling to do it in the same building. Postings put it plainly: "We train open frontier models and ship the same stack to our customers." The open-source output is real rather than decorative: PRIME-RL, Verifiers, PCCL and the full INTELLECT recipes are public, and research engineering roles list publishing at ICML and NeurIPS and writing technical blogs as part of the job.

Pay and equity are printed on the postings. Outsiders vouch for the execution speed: SemiAnalysis's ClusterMAX review in November 2025 said it went "from sending feedback in slack to a working cluster in a matter of hours" and praised the team's "relentless pace of shipping new features".

The headcount plan is explicit. The Head of Talent posting says the company will scale "from ~40 to ~100 people over the next twelve months, weighted heavily toward frontier research and infrastructure engineering", so most open roles are the first or second person on a problem: the first dedicated security hire, the first Head of Compute negotiating nine-figure GPU contracts, the first Head of Growth.

Prime Intellect describes its bar in terms of what you have built rather than where you worked. The open application asks for people who have "built complex technical systems, contributed meaningfully to open-source projects, or mastered multiple domains", and says "what matters is your ability to learn fast, think rigorously, and execute." Engineering postings repeat "We value potential over perfection" and ask for "a bias toward shipping, ownership of production code, and pragmatic engineering judgment".

The company is candid that the job is unstructured. Research postings ask for "comfort working in a fast-moving environment with ambiguous problems and high ownership", forward-deployed roles say "You don't wait for specs; you scope the problem, ship a solution, and iterate", and leadership roles are pitched as building a function from zero while reporting directly to the two founders. The stated mission language is consistent across every posting: open, decentralized AI, and the belief that model weights mean little if the infrastructure to improve them stays inside a few labs.

Prime Intellect hires through its Ashby board, where about 30 roles are open across Research, Applied Research, Engineering, Compute and Finance, Growth and Operations. The company does not publish its interview stages. What it asks for in writing is concrete: research engineering roles want hands-on PyTorch and distributed training work (FSDP, Megatron, DeepSpeed, vLLM, Ray), infrastructure roles want 3+ years running production GPU clusters with SLURM, Kubernetes and InfiniBand, and the open application asks you to describe "something impressive that you've built" and how you would accelerate open and decentralized AGI. For the RL research role the posting says that if you lack the listed experience but are a "high-energy person", read the linked resources and reach out anyway.

The Head of Talent posting says the company competes for candidates against offers from frontier labs and hyperscalers, and there is an open application for people who do not fit a listed role.