为什么值得看: Work with cofounder Adam Gleave at FAR AI to build research breakthroughs that make advanced AI systems trustworthy and secure.
中文摘要
FAR AI 致力于通过前沿研究提升先进 AI 系统的安全性与可信度。联合创始人 Adam Gleave 正在直接招聘工程师与研究负责人。岗位包含工程实现与科研探索方向。
创始人在 X 上的原帖
A
Adam Gleave
@ARGleave · 联合创始人, FAR AI
Come work with me to make AI turn out well! I'm hiring for research leaders, cracked engineers, exceptional project managers and operators. Our mission is simple: make advanced AI systems trustworthy and secure through research breakthroughs and driving industry-wide adoption. https://t.co/e2C7EgnOlg
FAR.AI is hiring Research Leaders, cracked Software/Research Engineers, Technical Project Managers, and Operators to build compute infrastructure, execute empirical red-teaming, and conduct frontier alignment research.
要做什么
Design, implement, and scale experimental pipelines to evaluate frontier models for deception, jailbreaks, and catastrophic risk.
Maintain and optimize high-performance GPU cluster infrastructure and experiment scheduling tooling.
Conduct adversarial red-teaming and pre-deployment testing for leading AI labs and public regulatory bodies.
Translate theoretical alignment approaches (interpretability, RL with deception probes) into high-velocity production code and reproducible benchmarks.
要求
Strong background in computer science, machine learning, distributed systems, or related technical disciplines.
Proficiency in Python and deep learning frameworks (e.g., PyTorch, JAX).
Demonstrated ability to move fast, run rigorous empirical experiments, or manage large GPU clusters.
Commitment to AI safety and understanding of LLM architectures and failure modes.
加分项
Prior publications in top ML conferences (NeurIPS, ICML, ICLR).
Experience with Kubernetes, Slurm, or large GPU training cluster orchestrations.
Previous red-teaming, prompt injection, or penetration testing experience.
整理自创始人的帖子和公司的公开页面。
条件
地点
Berkeley, CA / Singapore / Remote来自公司公开页面，帖子里没写
可远程
Yes (Global remote supported, or hybrid in Berkeley, CA / Singapore)来自公司公开页面，帖子里没写
薪资
$150,000 – $350,000 USD (Base salary varies by role: $150k–$275k for GPU Infra Engineers, $225k–$350k for Senior Research Engineers, up to $450k for Senior Research Leads). No equity as FAR.AI is a 501(c)(3) non-profit.来自公司公开页面，帖子里没写
Non-profit research institute making advanced AI systems trustworthy and secure.
FAR.AI conducts empirical AI alignment and safety research, focusing on adversarial robustness, mechanistic interpretability, and deception detection in frontier AI models. They develop automated red-teaming pipelines, conduct pre-deployment evaluations for frontier AI labs like OpenAI, and collaborate with regulatory bodies like the EU AI Office. Additionally, they maintain experiment-scaling infrastructure and run targeted grantmaking programs to expand the AI safety ecosystem.
客户: Frontier AI labs (e.g., OpenAI), government safety institutes (e.g., EU AI Office), and the broader AI research community.
Philanthropic Commitments / Grants · $30M+ · 2026-01 · Coefficient Giving (formerly Open Philanthropy), Schmidt Sciences, Survival and Flourishing Fund, Center for Security and Emerging Technology (CSET), AI Safety Fund (AISF)far.ai
团队成员
Adam Gleave · Co-founder & CEOPhD in Artificial Intelligence from UC Berkeley advised by Stuart Russell; former research intern at Google DeepMind working with Jan Leike and Geoffrey Irving; co-creator of Stable-Baselines3; board member at METR, LISA, and SAIF.gleave.me
Karl Berzins · Co-founder & PresidentCo-founder of FAR.AI leading organizational development and strategic operations.far.ai
Jenn Thom · Head of EngineeringEngineering leadership background managing technical infrastructure and computing clusters.far.ai
Boston Nyer · Chief Operating OfficerOperations leadership overseeing organizational growth, finance, and talent.far.ai
FAR.AI 是由 UC Berkeley 人工智能博士、前 Google DeepMind 研究员 Adam Gleave 于 2022 年联合创办的 AI 安全与对齐非营利研究机构（501(c)(3)）。机构专注于前沿大模型的对抗鲁棒性、欺骗检测和红队安全评估，曾为 OpenAI 等头部实验室和欧盟 AI 办公室提供测评服务。联合创始人兼 CEO Adam Gleave 在 X 平台发布招聘，招募“顶级工程师”（Cracked Engineers）、研究带头人、项目经理和运营人员。薪资待遇公开透明且极具竞争力（基础年薪约 15 万至 35 万美元，支持全球远程或加州 Berkeley 混合办公），并在 2026 年初获得了超 3000 万美元的多方资助。值得注意的是，由于是非营利组织，该机构不提供股票期权或股权激励。
我们查到的
FAR.AI is an AI safety research 501(c)(3) non-profit founded in 2022 by UC Berkeley AI PhD Adam Gleave and Karl Berzins.far.ai
In January 2026, FAR.AI announced over $30 million in funding commitments from Coefficient Giving, Schmidt Sciences, Survival and Flourishing Fund, CSET, and the AI Safety Fund.far.ai
According to IRS Form 990 / ProPublica records, FAR.AI brought in $24.3M in revenue in 2024 against $8.6M in expenses, providing exceptional runway.projects.propublica.org
The organization works directly with frontier AI labs (including OpenAI) and government institutions (such as the EU AI Office) on pre- and post-deployment safety evaluations.far.ai
As a 501(c)(3) non-profit organization, all positions are cash-compensated with no stock options or equity grants.far.ai
我们的预估
Stage
Non-profit Growth / Series A-B Equivalent Scale依据：Founded in 2022, >$30M in multi-year philanthropic commitments, 40-50 team members.
Team size
40–50 employees依据：Listed as 21-50 on 80,000 Hours and job boards cite scaling from 40 people.
Salary range
$150,000 – $350,000 USD依据：Explicitly published job posting salary ranges on 80,000 Hours and FAR.AI careers pages ($150k-$275k for Infra, $225k-$350k for Senior RE).
Equity range
0%依据：FAR.AI is a 501(c)(3) non-profit entity; equity compensation is legally not available.
Runway signal
3+ years (Extremely strong)依据：Secured $30M+ in committed funding in 2025/2026 with 2024 revenues of $24.3M against $8.6M expenses.
根据公开信息做的推断，未经公司确认。
可以问创始人的
Which engineering track has the most urgent hiring need right now (GPU Cluster Infrastructure vs. Red Team Engineering vs. Empirical Alignment Research)?
How are compute resources partitioned between exploratory internal research projects and contracted external evaluations for frontier labs?
What is the day-to-day balance between hands-on engineering work and co-authoring academic safety papers?