--- title: 'Member of Technical Staff, Evaluation Execution at METR' canonical: 'https://feeny.ai/job/member-of-technical-staff-evaluation-execution-metr-berkeley-rtt64jwna58y' type: 'job' last_seen: '2026-09-07' --- # Member of Technical Staff, Evaluation Execution at METR - **Company:** METR - **Location:** Berkeley, CA - **Compensation:** $328k–$579k - **Work type:** onsite - **Posted:** 2026-04-27 - **Last confirmed live:** 2026-09-07 - **Apply:** https://jobs.lever.co/metr/93baec4d-1e47-40f7-9990-6d7fef12da00 ## Job description ## About METR We are a nonprofit research organization that develops scientific methods to assess AI capabilities, risks, and mitigations, with a specific focus on threats related to AI R&D automation and misalignment. METR has consistently set precedents for catastrophic AI risk evaluations, including the first independent safety evaluations (working informally with Anthropic and OpenAI in 2022), the first loss-of-control evaluations and first agentic dangerous capability evaluations, the first evaluations using finetuning (mentioned briefly here), the first independent evaluations using internal information about training, the first review partnership for company risk analysis, the first embedded redteaming, and first evaluations of internal deployments, and the first independent misalignment incident investigation. We’ve been consulted and/or favorably referenced by groups on opposite ends of various spectra, including a16z, Khosla, Gary Marcus, Obama, and Dean Ball, and are known for producing one of the most positive results on AI capabilities (the time horizon trend) and the most negative (our downlift study). We’re generally referenced as the canonical third party assessor, e.g. as the obvious candidate to verify conditional pause agreements, and are trusted with AI incident investigations by frontierlabs and governments. We believe it is robustly good for policymakers and civil society to have a clear understanding of risks from AI systems, and we are extremely excited to build a team of ambitious, excellent people to tackle one of the most important challenges of our time. What this role looks like Running models on tasks. Often this means integrating models into our agent scaffolds, running them on our infrastructure and checking the results carefully. (METR both develops our own tasks internally and runs external evaluations.) Communicating results and takeaways. This includes designing useful graphs, writing up conclusions for different audiences (system cards, risk reports, regulators, X, etc), and having great takes on what matters for risk. Building software to improve our evaluations. We don't just try and run the same evaluation over and over again. We also run faster, more informative evaluations over time; this means making the right investments (with the support of our platform team). Project management. Live evaluations require keeping track of a bunch of threads and staying organized. With our recent risk report process, we were running many evaluations at once. Strong and professional communication. We run important and sensitive evaluations, and so the team needs to coordinate with METR leadership, lab contacts, regulators, and others. ## Why this role matters As part of informing the world about risk from frontier AI systems, METR often runs and publishes evaluations of frontier models. Our evaluations are a central tool the world uses to understand AI progress. Our Time Horizon methodology has been included in systemcards, called an "obsession" by the NYT, has wide reach online, and is used by governments to inform national policy. We’re expanding the ambition and scale of our evaluations. We have recently begun to measure model propensities and monitorability, and we are increasing the speed, reliability, and quantity of evaluations we aim to do so that we can keep the world informed. How METR’s evaluations are changing over 2026 Time Horizon is close to saturation, so we’re currently working on Time Horizon 2.0, which we expect to be running on models over the next 6 to 18 months. We’re gearing up for our first large-scale publication on monitorability, which we believe will be similar to TH in helping folks understand trends over time. We spent the past three months working on a large, industry-wide third-party risk assessment program - which includes us collecting information (and running evaluations!) for both monitorability and propensities/alignment. We expect to do much more work as part of our own risk assessment programs in the future. In general, many ambitious impact stories for METR require us having the capacity to run many more evaluations than we have run historically. For example, while our evaluations currently inform many key decisionmakers about AI capabilities, they are not yet consistently run with the scale, reliability, and speed necessary to play concrete, codified roles in regulatory frameworks. Unlocking this capacity is part of the near-future vision for evaluation execution. Required skills - Software engineering. You're a strong engineer with solid infra fundamentals. You can dig into unfamiliar systems, debug from logs, and identify and fix performance bottlenecks. - Speed and scrappiness. You get things done quickly. You’re able to quickly identify what 80/20 looks like, and then do that. - High attention to detail. You read closely, can spot bugs in transcripts, and pay attention to the important fiddly bits. Nice to haves - Research understanding and taste. You understand research ideas and priorities, and have good intuitions for which plots are informative and which analyses are worth running to poke at the data. - Strong external communicator. You communicate well with external stakeholders, and we trust you to stay on the ball with communications with, e.g., lab contacts. - Project management. You can juggle many balls at once, keep stakeholders updated, and track and anticipate blockers. - Strong writing ability. You can be a solid contributor to METR’s writeups of evaluation results, see e.g. our [GPT-5 report](https://metr.org/evaluations/gpt-5-report/#executive-summary). Our Culture METR is a mission-driven organization. We believe our work can meaningfully shape humanity's future for the better, and we want to be the best people in the world doing this work. We have a tight-knit, collaborative research culture rooted in truth-seeking and integrity. We're fiercely committed to producing high-quality, trustworthy science. We're honest and transparent about our results, especially when they may go against the grain. We've earned trust as reliable partners who handle confidential information with care. We maintain a low-ego, drama-free environment focused on what matters. Hybrid Preferred: Our technical team members are in our office in Berkeley 3-5 days/week. We would ideally like for you to be in person too, but we are happy to be flexible here. If you lack US work authorization and would like to work in-person, we can likely sponsor a cap-exempt H-1B visa for this role. We encourage you to apply even if your background may not seem like the perfect fit! We would rather review a larger pool of applications than risk missing out on a promising candidate for the position. We are committed to diversity and equal opportunity in all aspects of our hiring process. We do not discriminate on the basis of race, religion, national origin, gender, sexual orientation, age, marital status, veteran status, or disability status. We welcome and encourage all qualified candidates to apply for our open positions. ## About METR ## Company Overview - **One-liner**: METR (pronounced "meter") is a non-profit research organization that develops scientific methods to assess catastrophic risks from advanced AI systems by evaluating their autonomous capabilities. - **Entity Type**: Non-profit (funded by donations, not a typical private company; no equity) - **Headquarters**: Berkeley, California, USA - **Founded**: Not explicitly stated, but founded by Beth Barnes; likely ~2021–2022 - **Founders**: Beth Barnes (Founder, CEO) ## Core Business - **Primary industry**: AI safety research and evaluation - **Target customers**: AI developers (e.g., OpenAI, Anthropic, Google DeepMind, Meta, Amazon), governments, and the broader public - **Mission or purpose**: "Develop scientific methods to assess catastrophic risks stemming from AI systems’ autonomous capabilities and enable good decision-making about their development." ## Products & Services - **Frontier AI Capability Evaluations**: Systematic assessments of how autonomously AI systems can perform tasks (e.g., conducting research, developing apps, cyberattacks, self‑hardening). Published as open research. - **Risk Assessments & Safety Policies**: Advises AI developers and governments on risk assessment methodologies, including the "Responsible Scaling Policies" approach adopted by nine leading AI labs. - **Time‑Horizon Research**: Open‑source analysis showing that the length of tasks AI agents can complete doubles every ~7 months, a key input for forecasting transformative AI timelines. - **Evaluation Platform**: An open‑source platform built on Inspect AI for running AI agent evaluations at scale. - **Monitorability Evaluations**: Research on detecting AI agents attempting to evade monitoring or perform side tasks, including datasets of "reward hacking" and "sandbagging" behaviors. - **Productivity RCT**: A randomized controlled trial with experienced open‑source developers measuring how much AI tools actually boost productivity (finding systematic overestimation). ## Market Standing - **Valuation/Market Cap**: Not applicable (non‑profit) - **Key Metric**: Funding – METR is supported by donations from major foundations and individuals. Notable donors include The Audacious Project (TED), Jane Street, Sijbrandij Foundation, Pew Charitable Trusts, Schmidt Sciences, Packard Foundation, and others. Small part of income from a technical assistance contract with the European AI Office. **No funding from AI companies** (maintains independence). - **Notable Investors/Partners**: Partners with OpenAI, Anthropic, Google DeepMind, Meta, Amazon (pilot risk assessments); member of NIST AI Safety Institute Consortium, California Cybersecurity Task Force, UK AI Security Institute; technical assistance to European AI Office. - **Growth Signals**: Growing team (multiple open roles), expanding into cyberforensics and embedded assessments, research cited widely in AI safety policy, and adoption of their Responsible Scaling Policies by major developers. ## Competitive Advantages - **Independence**: No funding from AI companies, enabling unbiased, transparent research. - **Scientific Rigor**: Publishes all research openly; uses empirical methods (RCTs, time‑horizon analysis) rather than speculation. - **Policy Influence**: Their frameworks (e.g., Responsible Scaling Policies) are now industry standards, and they advise governments globally. - **Deep Technical Expertise**: Team of researchers and engineers building state‑of‑the‑art evaluations for autonomous capabilities, including security‑relevant evaluations. ## Strategic Focus - **Current priorities**: Developing methodologies to track AI loss‑of‑control risk, improving monitorability evaluations, expanding cyberforensics capabilities, and scaling the team to meet growing demand for independent evaluations. - **Direction for growth**: Increasing influence on AI governance, deepening partnerships with governments and companies, and building tools for continuous risk assessment. ## Why Work Here - **Mission‑driven**: Direct contribution to aligning AI development with public safety, working on one of the most critical problems of our time. - **Compensation**: Highly competitive with top AI labs (salary ranges $328K–$687K for technical staff, $150–$300/hr for contractors), plus benefits like medical/dental/vision, wellness ($1,500/yr), mental health ($6,000/yr), professional development ($5,250/yr), unlimited PTO. - **Work environment**: Small, fast‑moving, mission‑driven team based in Berkeley. On‑site preferred for technical roles (at least a few days a week), but hybrid and remote (including international) can be accommodated. Operations roles require in‑person. - **Hiring process**: Unique focus on work tests (1–3 take‑home tasks) and a 1–2 day paid work trial (travel and lodging reimbursed). Interviews are given substantially less weight. Commitment to diversity and equal opportunity. - **Culture**: Open, transparent, and empirical; values rigor and independence. Visa sponsorship available for technical roles. ## Sources 1. [metr.org](https://metr.org/) 2. [metr.org/about](https://metr.org/about#our-team) 3. [metr.org/careers](https://metr.org/careers) 4. [metr.org/hiring](https://metr.org/hiring) 5. [jobs.lever.co/metr](https://jobs.lever.co/metr) ## Other roles at METR - [General Counsel](https://feeny.ai/job/general-counsel-metr-berkeley-4s59mgfhjagg) — Berkeley, CA - [Member of Technical Staff, Cyberforensics](https://feeny.ai/job/member-of-technical-staff-cyberforensics-metr-berkeley-kxzz3emzx9gj) — Berkeley, CA - [System Administrator](https://feeny.ai/job/system-administrator-metr-berkeley-87zqbgq2ffsv) — Berkeley, CA - [Task Development Engineer](https://feeny.ai/job/task-development-engineer-metr-remote-5cbf872b6d7w) - [Member of Technical Staff, Embedded Assessments](https://feeny.ai/job/member-of-technical-staff-embedded-assessments-metr-berkeley-jsprvy2fdmqy) — Berkeley, CA - [Member of Technical Staff, Security Engineering](https://feeny.ai/job/member-of-technical-staff-security-engineering-metr-berkeley-w44zdwnsrdma) — Berkeley, CA - [General Expression of Interest](https://feeny.ai/job/general-expression-of-interest-metr-berkeley-avs3pft13sf0) — Berkeley, CA