FREE ACCESS
5,000–10,000 jobs/day
See all jobs on Scoutfield
Search thousands of fresh jobs every day.
Discover
- Fresh listings
- Fast filters
- No subscription required
Create a free account and start exploring right away.
Core Competencies
Role fitCore Competencies
Use this summary to align your resume positioning with the role.
Demonstrates expertise in post-training model development and deception detection in AI, with strong capabilities in Python and PyTorch. Proven ability to mentor technical staff and collaborate effectively with research teams to enhance model reliability and safety.
Highest-signal resume keywords
Post-Training Model DevelopmentPython ProgrammingPyTorch FrameworkResearch JudgmentAI Safety Research
ATS Keywords
Tailor your resumeApplicant Tracking System Keywords
Tip: use these terms in your resume and cover letter to boost ATS matches.
Hard Skills
Post-Training ImplementationModel TrainingReinforcement LearningFine-TuningLarge Model ManagementWhite-Box MethodsDeception DetectionExperimental DesignData AnalysisModel Evaluation
Soft Skills
MentoringCollaborationCommunication
Tools & Technologies
FAR.AI Compute ClusterResearch Codebases
Industry Keywords
Frontier AIAI AlignmentInterpretabilityResearch PublicationsModel Organisms
Tech Stack
Tools & technologiesPythonPyTorch
About the role
Key responsibilities & impact- Lead research projects on white-box deception detection, probe-based monitoring, and training against probes
- Define research questions, set milestones, choose experiments, and assess result quality
- Run post-training and RL experiments on large models, including RLVR and fine-tuning with probe-based penalties
- Diagnose training instability, reward hacking, and infrastructure failures
- Design model organisms that stress-test methods against realistic failure cases
- Mentor Members of Technical Staff on research judgment and experimental rigor
- Partner with the infrastructure team to plan and run large post-training jobs on FAR.AI’s compute cluster
- Enable frontier labs to adopt the team’s methods through reproducible code, clear evaluations, practical guidance, and direct collaboration with lab safety teams
- Write and present findings for research papers, frontier lab partners, and the broader AI safety community
- Develop white-box probes to identify deception from internal activations and assess failure modes
- Turn probes into real-time monitors with low false-positive rates, minimal overhead, and reliability across tasks and model versions
- Use probes as training signals and test whether models become more honest or better at concealing deception
Requirements
What you’ll need- 3+ years experience implementing post-training, model training, or closely related efforts in a frontier AI environment
- Hands-on experience post-training models at scale, including RL, RLHF/RLVR, or large fine-tuning runs
- Ideally experience with models of 400B+ parameters
- Strong Python and PyTorch skills
- Experience working in large, shared research codebases
- Strong research judgment
- Based in or willing to work in U.S. time zones
- Experience with safety or alignment research at a frontier lab or research lab working on large models is an additional advantage
- Experience with interpretability or white-box methods is an additional advantage
- Experience building model organisms of misalignment, or studying reward hacking, scheming, or evaluation awareness is an additional advantage
- Publications at top ML venues or widely cited technical reports are an additional advantage
- A PhD in a relevant field is an additional skill they are excited about, not a stated requirement
Benefits
Comp & perks- Full-time employment (40 hours/week)
- Remote work within U.S. time zones
- In-person work in Berkeley, California
- Visa sponsorship for in-person employees
- Employee status through FAR.AI in the USA or Singapore
- EOR employment or contractor arrangement outside the USA or Singapore
- Paid work trial lasting up to 1 week