Researcher, Evaluations and Benchmarks

Job not on LinkedIn

🔥 0 minutes ago

🏄 California – Remote

🏄 California – Remote

⏰ Full Time

🟡 Mid-level

🟠 Senior

👻 Ghost score 10%

What it means: how likely this posting is sitting open without active hiring.How to read it: 60% or higher means we think this is a ghost. Lower scores are less of a worry.How we score it: age, reposts, open roles vs team size, how fast jobs close, and whether pay is listed.Fine print: an estimate from public data, not a judgment of the company. Real jobs can score high, so treat it as a hint, not a fact.

Apply Now

Find Similar Remote Jobs

Set Up Email Alert For New Jobs

📊 Check your resume score for this job

Improve your chances of getting an interview by checking your resume score before you apply.

Upload Resume

ActiveFence

WebsiteLinkedInAll Job Openings

201 - 500 employees

🔒 Cybersecurity

🤖 Artificial Intelligence

💰 $400M Series B on 2021-07

Cybersecurity • Artificial Intelligence

ActiveFence is a Trust and Safety provider for online platforms, protecting platforms and their users from malicious behavior and content. Trust and Safety teams of all sizes rely on ActiveFence to keep their users safe from the widest spectrum of online harms, unwanted content, and malicious behavior, including child safety and exploitation, disinformation, hate speech, terror, nudity, fraud, and more. We offer a full stack of capabilities with our deep intelligence research, AI-driven harmful content detection, and online content moderation platform. Protecting over three billion users globally everyday in over 100 languages, ActiveFence lets people interact and thrive online.

📋 Description

• Ship a benchmark every two to three weeks measuring a previously unmeasured frontier risk • Collaborate with leading AI labs and universities on benchmarks and papers • Own benchmark taxonomy, harness, quality bar, and release • Read evals personally and verify taxonomy alignment, rubrics, verifiers, and data distribution • Hold the benchmark plan and calendar and keep researchers on timeline • Direct two or three freelancers/SMEs as needed • Meet roughly monthly with the CTO, pod, and research leads to revise the quarterly release plan • Tie the release roadmap to target accounts • Spend approximately 20% of time monitoring the ecosystem and reading research • Maintain contacts inside AI labs and speak with lab personnel weekly • Travel to conferences a couple of times a year

🎯 Requirements

• PhD or Masters in computer science, machine learning or a related field, or equivalent depth from industry research • 3+ years building and running safety or security evaluations for language models in production, at an AI lab, a model provider, or a safety and security research organisation • 5+ relevant research publications in AI safety and security, including lead author on at least 2 • Strong engineering skills, including evaluation harnesses, distributed inference, vLLM, reading a codebase and fixing it • Ability to build a taxonomy, not only score against one • Ability to direct a researcher and two freelancers without formally managing them • Strong English, written and spoken • Curiosity about AI harms and ability to learn a new subject every three weeks • Ideally: post-training experience with SFT, DPO, or GRPO • Ideally: agentic evaluation experience including tool use, orchestration, permissions, and prompt injection • Ideally: publications at top conferences • Willingness to present own work on client calls • Strong verbal and written communication and ability to present to large and/or senior audiences • Willingness to travel to conferences at least 3 times a year

🏖️ Benefits

• Budget for freelancers (SMEs) directed ad hoc when needed • Travel to conferences at least 3 times a year • Opportunity to collaborate with leading AI labs and universities • Access to approximately 150 researchers working on harms • Conference travel a couple of times a year / at least 3 times a year