How to Apply to Nous Research

6 min read Last updated April 21, 2026 7 open positions
Blake Crosley
Blake Crosley Researched using publicly available information, hiring data, and ATS analysis.

Key Takeaways

  • Apply directly at nousresearch.com/careers — no third-party ATS, no recruiter intermediary
  • The free-text 'why' field matters more than the resume; reference specific Nous work
  • Open-source artifacts (GitHub, HuggingFace, papers) outweigh job titles at any company
  • Expect 45-60 minute technical deep-dives with a founder or senior researcher early in the process
  • Nous optimizes for open-source researchers, not hired-gun engineers — alignment on philosophy matters
  • Interviews are short (3-5 rounds total) and turn around quickly — 2-4 weeks end to end is typical
  • If you're more motivated by compensation than by shipping public research artifacts, larger labs will likely be a better fit

About Nous Research

Nous Research is an AI accelerator company focused on building open-source foundation models and decentralized training infrastructure. The lab is best known for the Hermes model series — a family of open-weights language models fine-tuned for steerability, truthful instruction following, and agentic reasoning — and for DisTrO, a distributed optimizer that makes it feasible to train large models across geographically dispersed GPUs over the public internet. Nous also operates Psyche, a decentralized training network that turns DisTrO into a production system that community contributors can participate in. The founding team includes Karan Malhotra, Teknium (Michael Emberson), Shivani Mitra, and Bowen Peng; the company is headquartered remote-first in the United States and raised a Series A led by Paradigm in 2025. The team is small and research-dense: Nous is not a product company in the SaaS sense — it ships models, papers, and open-source infrastructure. That shapes everything about how they hire.

Application Process

  1. 1
    Review the open roles at nousresearch.com/careers

    Nous posts a small, curated set of roles — typically fewer than ten at any given time — directly on their WordPress careers page. Each role links to a dedicated page with a description and an application form. There is no large ATS (no Greenhouse, no Lever). If you don't see an exact match, there is no general 'apply to pool' form — the expectation is you apply to a specific role.

  2. 2
    Prepare a concrete artifact portfolio before you apply

    For ML engineer and research scientist roles, the Nous team cares more about code and models you've shipped than your resume. Collect GitHub links, arXiv papers, HuggingFace model cards, training runs, and open-source PRs into a single public URL (personal site, gist, or GitHub README). The most successful applicants have at least one publicly verifiable contribution to foundation model training, distributed systems, RL, or evaluation — ideally tied to a system running at meaningful scale.

  3. 3
    Submit the application on the role page

    Open the role's dedicated page at nousresearch.com/{role-slug}/ and complete the application form. It typically asks for your name, contact, resume, and a free-text field for why you want to work on this specific problem. Use the free-text field — that's where most applicants get filtered in or out. Be specific about the Nous work you've followed (e.g., Hermes 4, DisTrO, Psyche) and what you'd want to build.

  4. 4
    Expect a technical conversation with a founder or senior researcher

    Because the team is small, first-round interviews are typically conducted by one of the founders or a senior researcher on the relevant sub-team. Expect a focused 45-60 minute conversation about a project on your resume — they will drill into the details. This is not a generic 'behavioral' round; they want to see that you actually built what you claim and understand the failure modes.

  5. 5
    Work sample or paired session

    Technical rounds often involve reviewing a real piece of Nous infrastructure or a small open-ended problem — e.g., debugging a training run, sketching a data pipeline, or proposing an evaluation. For research roles, expect a discussion of a paper you've read recently in depth. There is less emphasis on timed LeetCode-style interviews than at larger labs.

  6. 6
    Founder conversation and offer

    The final step is typically a conversation with a founder about fit, thesis, and long-term direction. Nous is open-source-first and research-dense, so alignment on that philosophy matters. Offers follow within a week or two of the final round.

Resume Tips for Nous Research

critical

Lead with open-source contributions, not job titles

A GitHub handle with a long training-systems commit history beats a big-tech title on the first page. List your most-starred repos, papers, and model releases near the top. If you have a HuggingFace profile with published weights or evals, link it prominently.

critical

Name specific systems you've worked at scale

Generic 'trained large models' is too vague. State the model size, token count, GPU count, parallelism scheme (FSDP, DeepSpeed, Megatron, custom), and throughput numbers. For RL, state the training algorithm, environment, and reward model details. Concrete numbers signal credibility.

recommended

Show evidence of debugging training runs, not just launching them

MLE roles at Nous heavily favor candidates who've recovered runs from divergence, diagnosed loss spikes, profiled throughput bottlenecks, or chased down dataset contamination. Bullet these experiences specifically.

recommended

Cite Nous work you've used or followed in your cover note

The free-text application field is where you make yourself memorable. Reference specific Nous papers or models you've read or used — Hermes 4, DisTrO, YaRN, or one of their evaluation datasets. Say what you'd build next.

recommended

For Research Scientist: include a citation list

Published or pre-print research with citations matters more than years of experience. Include a short list of your most-cited papers with venue and year. If you've written widely-read technical blog posts that drove community adoption of a technique, include those too — Nous values public-facing research output.

nice_to_have

Keep it to one page

The team is small and reads applications closely. Dense one-page resumes with linked evidence outperform multi-page resumes full of bullet points. If you need more space, put it on a personal site and link it.

ATS System: Custom WordPress careers page

Nous does not use a conventional ATS like Greenhouse, Lever, or Workday. Their careers page is a custom-built section of nousresearch.com (WordPress/Elementor), with each role having its own page and application form. There is no public API for their applicant tracking, no branded 'apply with LinkedIn' button, and no automated keyword screening. Your application is read by humans — usually the hiring manager for that role — within days of submission.

  • No ATS keyword stuffing needed — a human reads every application
  • Apply to the specific role page, not a general inbox
  • The free-text 'why' field is the most-read part of the application
  • Include direct links to artifacts (GitHub, papers, models) — don't expect recruiters to search for you
  • Response times are faster than at large labs — typically 1-2 weeks, not 4-8

Complete Custom WordPress careers page Resume Guide →

Interview Culture

Nous interviews are technically deep and culture-forward.

First rounds are typically a 45-60 minute conversation with a founder or senior researcher, conducted over video call. The conversation focuses on specific projects on your resume — expect to be asked what would have broken if you'd done X differently, why a particular design decision was made, and how you knew the solution was correct. There are fewer whiteboard coding rounds than at large labs, but more 'walk me through this paper' and 'how would you debug this training divergence' style questions. The culture is research-lab-like rather than big-tech: small team, high autonomy, strong opinions on open-source and decentralized systems. Candidates who have spent time on Twitter/X following the AI research community, who've read and can discuss recent papers, and who have public opinions on open-weights vs closed-weights debates tend to resonate. The team appreciates candidates who push back thoughtfully during the interview — agreeable-but-shallow answers do not land well. Compensation discussions happen late in the process and are typically a mix of base salary, equity, and — notably — meaningful participation in the open-source ecosystem Nous is building. For candidates motivated primarily by a public track record of research and shipped models, this is attractive. For candidates optimizing purely for base pay, larger labs will typically pay more.

What Nous Research Looks For

  • Demonstrated ability to train large models — stated with specific parameters, token counts, and hardware
  • Public artifacts: GitHub commits, papers, model releases, or technical blog posts with real-world traction
  • Fluency in the open-source AI research community — you read the papers, you have opinions
  • Experience with distributed systems, especially anything involving GPU clusters or parallel training
  • A research mindset — willing to run experiments that might fail and report results honestly
  • Strong writing: your application, your code comments, your model cards — Nous ships documents as much as code
  • Autonomy — the team is small and ownership is high; candidates who need heavy management are a poor fit

Frequently Asked Questions

Does Nous Research hire remotely?
Yes. Nous is remote-first and hires globally for most roles, with the caveat that US-based time zones are preferred for roles that require heavy synchronous collaboration with the core team. The careers page lists most roles as remote without a specific location requirement.
Do I need a PhD to work at Nous Research?
No. Nous hires based on shipped work and research taste, not credentials. For Research Scientist roles, a PhD or demonstrated peer-reviewed research output is common but not required — a strong independent track record of papers, models, or widely-used open-source tools is equally valued. For MLE roles, a degree is not required at all.
What programming languages and frameworks should I know?
Python is baseline. For training infrastructure roles, deep knowledge of PyTorch, distributed training frameworks (FSDP, DeepSpeed, Megatron-LM, or custom equivalents), and CUDA awareness are expected. For RL work, familiarity with vLLM, TRL, Axolotl, or similar open-source training libraries is a strong signal. For infrastructure roles, understanding of NCCL, InfiniBand, and cluster orchestration (Slurm, Kubernetes) is useful.
How long does the Nous Research interview process take?
Typically 2-4 weeks from initial application to offer, which is faster than most large AI labs. The process usually includes: resume review (1-5 days), a technical conversation with a founder or senior researcher (45-60 min), a work-sample or paired session (1-2 hours), and a final founder conversation. Response times are notably faster than at large labs because the hiring manager is typically one of the founders, not a recruiter.
What's it like to work at Nous Research?
Small team, research-dense, open-source-first. Engineers have a high degree of autonomy over their work and contribute directly to public releases — Hermes model weights, DisTrO papers, Psyche network code. There is no 'internal-only' product work; everything Nous ships is visible to the outside world. This appeals to people who want their work to be publicly credited and discussed; it can feel intense for people who prefer stable, well-scoped projects with clear handoffs.
Does Nous sponsor work visas?
Nous is small enough that visa sponsorship is decided case-by-case rather than as a blanket policy. For senior research or MLE hires, the team has historically been willing to sponsor; for more junior roles, it's less common. If visa sponsorship is required, mention it in your application so the team can assess fit early rather than late in the process.
What should I include in the free-text 'why' field on the application?
Be specific about (1) which Nous work you've followed or used — name a model, paper, or project; (2) what technical problem you'd want to work on; (3) what makes you credible for that problem — one or two concrete artifacts or past experiences. Avoid generic enthusiasm ('I love open-source AI') and avoid reciting the job description back. The team reads every word, and specificity is the single strongest signal.
What's the compensation like?
Nous pays competitively for a small, VC-backed AI lab — base salary plus equity, with the equity component being meaningful given the company's growth trajectory and the Paradigm-led Series A at a ~$1B reported valuation. Base salaries are typically lower than at Anthropic, OpenAI, or large FAANG AI orgs, but the equity upside is larger, and compensation includes the less-tangible benefit of shipping public research artifacts under your name. Specific ranges are discussed late in the interview process.

Open Positions

Nous Research currently has 7 open positions.

Check Your Resume Before Applying → View 7 open positions at Nous Research

Related Resources

Similar Companies

Related Articles

Sources

  1. —
  2. —
  3. —
  4. —