Author: @hiiinternet
Compiled by: Peggy
Editor's Note: Anthropic is often imagined as an AI lab comprised of PhDs, researchers, and cutting-edge model experts, but this breakdown of the resumes of 1,680 engineers provides a more realistic answer: Anthropic's core is not just "research," but "construction."
This article analyzes 5,306 LinkedIn profiles of people currently employed by Anthropic, and further filters out the resumes of 1,680 engineers, arriving at a counterintuitive conclusion: Anthropic's core talent profile is not the "researchers" that outsiders imagine, but a group of experienced "builders" (people who can truly build, run, and scale large-scale systems).
Data shows that Anthropic's engineering team has grown rapidly in the past 18 months: more than half of the engineers have been with the company for less than a year, but the new employees are generally very senior, with a median of 12.2 years of work experience before joining, and many come from companies known for their engineering capabilities and infrastructure, such as Google, Meta, Amazon, Microsoft, Stripe, Databricks, Snowflake, and Palantir.
This also explains the true focus of Anthropic's engineering organization: compared to the model research that attracts external attention, it is more like a highly engineering-focused infrastructure company. Its engineers mainly have backgrounds in infrastructure, backend, distributed systems, databases, and security; only 13.7% hold doctoral degrees, with the majority being senior engineers with bachelor's or master's degrees.
Early-career professionals are not entirely without opportunities, but the threshold is extremely high: internships at top tech companies, competition results, paper publications, or AI security/alignment project experience often become screening signals for the replacement of years of work experience.
The author's final advice is straightforward: if you want to join Anthropic, don't write your resume like you're applying to a research lab; instead, highlight the large-scale systems you've actually built, expanded, and maintained. At the core of the competition in cutting-edge AI, it's increasingly becoming a competition of engineering and infrastructure capabilities.
The following is the original text:
Builders, not researchers
I scraped all 5,306 LinkedIn profiles that listed Anthropic as their current employer. I then filtered out 1,680 of them who were actually in engineering roles and further examined 7,986 records in their past job descriptions to analyze what they were doing before joining Anthropic.
The following are the results.
The organization expanded almost overnight.
Only 15 engineers joined Anthropic before 2021 and are still employed there. In 2025, the organization's engineering team nearly tripled, adding 686 engineers that year; the hiring pace in 2026 is expected to be similar, with 455 new employees added as of June.
Currently, half of the engineering team has been with Anthropic for less than a year. 53% joined within the past 12 months. Median tenure: 10 months.
This is a large organization, yet it was built in approximately 18 months.
They almost exclusively hire senior engineers.
The median prior work experience at Anthropic was 12.2 years. The middle 50% had between 8.8 and 16.5 years of experience. Of the 1,680 people, only 50 had less than 3 years of experience. 44% had 13 years or more of experience. Recruiting recent graduates was virtually nonexistent.
In other words, a typical new employee at Anthropic is an engineer with 12 years of experience who has only been with Anthropic for 10 months.
It clearly leans more towards infrastructure than traditional research.
Infrastructure backgrounds appeared in 40% of engineers' resumes. Backend, distributed systems, databases, and security each accounted for about 20%. Reinforcement learning, the "RL" in RLHF, appeared in only 3.3% of resumes.
A typical Anthropic engineer has spent the last decade building large-scale production systems for a hyperscale cloud vendor or a heavy infrastructure startup.
Their self-listed skills also illustrate the same point: 585 people with Python skills, 566 with Java, 443 with C++, 376 with JavaScript, 302 with SQL, 230 with Linux, 189 with distributed systems, and 154 with AWS. While the more "sexy" work of model training certainly exists, it represents a very small percentage.
The biggest source of talent is not laboratories, but Google.
Everyone assumes Anthropic primarily poachs talent from OpenAI and DeepMind. However, its largest talent pipeline, far ahead of the competition, is Google's. Those rival labs are merely two small bars in the middle of the chart.
Anthropic clearly prefers companies known for their engineering rigor: Stripe, Databricks, Snowflake, Palantir, and Airbnb.
Looking at where these engineers have worked historically, the ranking is as follows: Google 405, Meta 273, Amazon 197, Microsoft 171, Stripe 124, Apple 87, Stanford 68, DeepMind 62, Airbnb 51, and OpenAI 48. Currently, half of the engineering teams, or 50%, have at least one FAANG company listed on their resumes.
Of course, they are also poaching talent from other AI labs. OpenAI is one of the top five direct sources, and DeepMind is one of the top six. Approximately 94 engineers have joined Anthropic directly from other cutting-edge AI labs.
Myths about PhDs
Only 13.7% of people have a doctoral degree. That's about one in seven.
Anthropic typically recruits senior engineers with bachelor's or master's degrees, not research scientists. The notion that "the entire lab is full of PhDs" is fundamentally flawed at the engineering team level.
The distribution of academic backgrounds perfectly matches the profile of a "constructive organization": 819 people have backgrounds in computer science, followed by 78 in mathematics, 70 in physics, and 69 in computer engineering. Philosophy also made it into the top 20 with 13 people, which may be related to the security field.
Stanford is clearly ahead in terms of recruitment sources.
Looking at the schools, the historical cumulative rankings are as follows: Stanford 144, Berkeley 118, MIT 80, CMU 73, Harvard 42, Cambridge 39, UW 36, Waterloo and Cornell 35 each, Oxford 33, and Princeton 32. The first four schools combined account for a quarter of the entire engineering team.
80% of the people have the same job title.
"Member of Technical Staff"
A former Instagram CTO, several former Adept founders, and Stanford faculty members all have the title "MoTS" at Anthropic. This flattening of job titles is clearly intentional. Qualifications and specific functions are hidden in the design.
For those in the early stages of their careers, where is the only pathway to Anthropic?
172 engineers have less than 6 years of work experience, including 50 with less than 3 years. However, they are not recent graduates in the conventional sense. They are roughly divided into two categories, with almost no ordinary mid-level engineers in between.
Compared to the entire engineering team, they exhibit distinct characteristics: a higher percentage of PhDs, reaching 19%, compared to 13.7% overall; a three-fold increase in product/SWE titles, reaching 15%, compared to only 5% overall; and a significantly lower probability of having FAANG backgrounds, at only 32%, compared to 50% overall.
What they have in place of years of service is another kind of prestige:
Internship opportunities. 50% of respondents listed internships at the following companies: Meta (16), Google (10), DeepMind (6), Microsoft (5), Amazon (5), as well as Jane Street, Two Sigma, HRT, Optiver, and Nvidia.
From quantitative trading to AI labs. 9% of them have worked at top trading firms, including Jane Street, Two Sigma, Five Rings, HRT, Optiver, and Citadel. This is a group of young, math/computer science competition-caliber talents who have entered the AI lab through the high-frequency trading industry.
Alignment with Fellowship. 6% of people have had contact with MATS, SERI, Redwood, or ARC. This is an entry point that is almost exclusively open to early-career talent and virtually non-existent among senior members.
A very clear profile is: MIT graduate, IOI silver medalist, Codeforces score of 2900+, who directly entered the field of reinforcement learning and security after four years of work experience. Their selection criteria are not years of work experience, but rather competition rankings and paper publications.
These young engineers are also more international than their senior counterparts. The academic backgrounds of the junior engineers include: Berkeley (15), Stanford (14), Cambridge (10), MIT (7), Tsinghua University (7), Oxford University (6), as well as Imperial University, NUS, Shanghai Jiao Tong University, and ETH Zurich.
So, how should you interpret this information?
If you want to join Anthropic as an engineer, don't write your resume like you're applying to a research lab; write it like you're applying to an infrastructure company. Showcase systems you've actually built and scaled. That's the kind of resume that gets you hired.
The early career stage is the only exception. At this stage, the threshold is not ordinary work experience, but top internships, competition rankings, or academic papers.
If you're competing with Anthropic for talent, your target isn't "PhDs" or "laboratory backgrounds" themselves, but rather seasoned builders from hyperscale cloud vendors or companies with strong engineering reputations: they have around 12 years of experience and might come from Stripe, Databricks, Snowflake, or Palantir. Anthropic is already aggressively recruiting from this talent pool.
