Nuro AI Labs is a small London lab shipping frontier research and commercial products in parallel. AVALON-2B is open-source. Hypersave is in production. Khyaa is in the hands of US financial advisors. Every hire owns something material on day one.
No sub-sub-team layers. The person who designs the system trains the model, tunes the prompts, writes the docs, and watches the latency graph. You will be named on releases, papers and product launches.
AVALON-2B is Apache 2.0 with weights, code and a reproducible benchmark suite. PLMR is heading to preprint. Hypersave runs in production at sub-200ms p95. The work is real, in the open, and measured against the frontier.
No coding gauntlets, no whiteboarded riddles, no twelve-loop interviews. A conversation, a paid one-week work sample, a written decision in five working days. We give feedback either way.
Two weeks end-to-end for most candidates. We pay for your time on the work sample — it is real, shippable, and the same brief a teammate would receive.
Email careers@nuroailabs.com with the role in the subject line, your CV or GitHub, and one short paragraph on why this role specifically. No cover letters.
A 30-minute call with Akhil or Naga. Mutual screen — what you've built, what you want to build, whether the lab is the right shape for you.
A paid one-week trial on something real and shippable. Same brief we'd hand a teammate. We pay your standard rate.
A yes or no within five working days of the trial wrapping. We give written feedback either way.
If you join, you're signing up for these. Read the long-form versions on /about.
We build AI for a living. It would be strange to ask candidates to pretend AI tools don't exist. You are welcome — encouraged, even — to use AVALON, Claude, GPT, Cursor or anything else throughout the application and the take-home portion of the work sample.
The final stage of the work sample is in-person (London) or live on a video call. We sit together with the artefact you produced and walk through the design choices, the failure modes, the rejected alternatives. The bar isn't whether you used AI; the bar is whether you can defend the result and extend it on the spot.
If a candidate's final-stage answers cannot account for the artefact they submitted, we treat that as a no — same as we would for any other source of unattributed work.
Yes — three- to six-month research internships, primarily for MSc / PhD students working on memory, retrieval, byte-level architectures or agent systems. Apply via the same address with 'Internship' in the subject and a short note on what you'd want to ship with us.
Both. Most roles are London-first with a relaxed in-office expectation, or fully remote within ±5h of UK time. GTM roles in NYC are in-person. Khyaa product engineering needs US/EST overlap because the design partners are East Coast RIAs.
Yes. We commit to a written response on every application within ten working days, and substantive feedback after any work sample. The bar is high; the process is honest.
We sponsor UK Skilled Worker visas for senior research and founding engineering roles where the impact justifies it. We do not currently sponsor US work authorisation.
Please do — after twelve months, or sooner if you've shipped something material. We track every application and read re-applications in the context of what you've built since.
Yes. If you think you'd compound the lab, write to careers@nuroailabs.com with what you'd own. We've hired into roles we never advertised.
Nuro AI Labs does not work with external recruitment agencies. We do not ask candidates for payment, equipment deposits, banking details or personal documents at any stage. We do not interview over WhatsApp, Telegram or Signal. If someone claiming to represent Nuro contacts you from a Gmail, Outlook, or any non-@nuroailabs.com address, it is not us — please report it to careers@nuroailabs.com.
One slow email per release. New papers, new model drops, new openings — before they hit the public channels.
