The Anthropic Fellows Program is a paid, four-month, full-time research fellowship that pays roughly $3,850 per week, plus benefits and about $15,000 a month in compute. No formal degree is required, it spans five tracks, and in past cohorts more than 80% of fellows produced a public research paper.
What is the Anthropic Fellows Program — and what does it actually pay?
Anthropic just opened applications for its Fellows Program, and most people have the wrong idea about who it's for. Here's the actual deal: four months, full-time, roughly $3,850 per week — plus benefits and around $15,000 a month in compute funding to run your experiments. You work on a real research project alongside a mentor, and the goal is a public output, usually a paper.
Here's the line most people skim right past: the program is open regardless of previous experience, and the formal degree requirement does not apply. Read that again. You do not need a Ph.D. You do not need to already work in AI. Applications are rolling, and the next cohort starts this fall.
Which of the five Anthropic Fellows tracks is right for you?
There are five tracks — and they are not all for engineers. Three are technical and expect real engineering chops. Two are open-door tracks built for people coming from non-traditional backgrounds.
The 3 technical tracks (real engineering chops needed)
- ML Systems & Performance — building simulators, accelerator backends, and data pipelines. Mentors include Alwin Peng and Zygi Straznickas.
- Reinforcement Learning — RL environments, algorithms, and training-data quality. Mentors include David Brandfonbrener, Kaidi Cao, and Will Williams.
- AI Security — pentesting, vulnerability research, and red-teaming. Mentors include Nicholas Carlini, Keane Lucas, and Newton Cheng.
The 2 open-door tracks (non-traditional backgrounds welcome)
- AI Safety — interpretability, scalable oversight, and AI welfare. Built for people transitioning into research. Mentors include Sam Bowman, Alex Tamkin, Ethan Perez, and Kyle Fish.
- Economics & Policy (The Anthropic Institute) — AI's effects on work, society, and policy. Prior experience is "a plus but not required." Mentors include Jack Clark, Maxim Massenkoff, and Marina Favaro.
The two open-door tracks are where most people overlook their own shot. AI Safety explicitly welcomes career-changers, and the Economics & Policy track treats prior experience as "a plus but not required." If you've been telling yourself you need to be a machine-learning engineer to work at a frontier lab, these two tracks are Anthropic saying otherwise — out loud, in the posting. Pick the track that matches how you already think, not the one that sounds most impressive.
Do you need to know Python — and how should you apply?
Yes, the posting lists Python as a requirement. Don't let that stop you. Anthropic literally encourages people to apply even if you don't believe you meet every single qualification — and notes that people from underrepresented groups are especially prone to imposter syndrome and to talking themselves out of applying. Read that twice. They are telling you to apply anyway.
And Python is learnable fast. Either of these free courses gets you functional enough to stop counting yourself out:
- Google's Python Class (free) — built for people who already know how to think logically. It's organized as an intensive two-day class with written materials, lecture videos, and lots of coding exercises. 👉🏾 developers.google.com/edu/python
- Scrimba's Learn Python (free) — if you want something interactive, it's about 5.6 hours and lets you edit the instructor's code right inside the lessons. 👉🏾 scrimba.com/learn-python-c03
If I were applying, here's my exact strategy
Most people stop at "I'll submit my resume." The fellows who stand out study the people and the work before they apply — and AI makes that fast. Two prompts do the heavy lifting. Run them in the same chat, back to back. One to understand them, one to position you.
Research the mentors for the [track name] workstream of the Anthropic Fellows Program: [list the names from your track]. Look up their recent papers, LinkedIn, and blog posts. Then analyze what these people care about most right now, the problems they're trying to solve, the themes that connect their work, and what kind of candidate or experience would catch their attention. Be specific and concrete.
Based on that analysis, here's my current resume and background: [paste]. Act as an expert AI research recruiter and interview me one question at a time. Don't limit yourself to what's on my resume — ask about projects, skills, side work, coursework, and experiences that might not be written down anywhere, so we surface anything in my background that's relevant to what these mentors care about. Keep going until you're confident we haven't missed something. When you're done, rewrite my resume and make sure we accentuate my skills that apply, so that nothing that can help me is left out.
Apply here
👉🏾 Submit your application: Anthropic Fellows Program application form
👉🏾 Read the full posting: Program details & requirements
Rolling admissions. Next cohort this fall. Don't count yourself out before you start.
The Anthropic Fellows Program pays you to break into AI research — and explicitly invites people without a traditional background to apply. Two of the five tracks welcome career-changers, Python is learnable in a weekend, and the application rewards anyone willing to study the mentors first. The only disqualifying move is not applying.