Graduate Research Scientist: AI Safety & Frontier Models

ByteDance; San Jose (CA); On-site; USD 218,000 - 388,000; Full time

Benefits offered by this job

401(k) with company match

Medical, dental, vision insurance

Parental leave

Disability coverage

Wellbeing benefits

Paid holidays

Paid sick days

Paid Personal Time

Job summary

ByteDance in San Jose is seeking a Research Scientist Graduate (Seed Model‑Alignment & Preparedness) for a 2027 start. You will join the Technology team as a regular, full‑time hire to push forward alignment, safety evaluation, and preparedness for frontier models and agent-enabled systems.

As a new graduate, you will tackle frontier challenges, design agent RL experiments, and contribute to automated red‑teaming, monitoring, and failure analysis.

Qualifications

• Minimum qualifications include BS in CS/EE/Math/Physics or related discipline; strong ML foundations and RL exposure.
• Preferred qualifications include MS in CS/EE/Math/Physics or related discipline; experience with SFT/RLHF and research in AI safety or scalable oversight.

Responsibilities

• Agent RL & safety training: design environments, rewards, data to reduce deception and unauthorized actions; ensure generalization to new tasks.
• Frontier risk assessment: evaluate dangerous capabilities, model threats, build safeguards with empirical evidence.
• Automated red teaming: develop automated attack generation and adversarial training to harden models.
• Monitoring & oversight: build oversight models, check permissions, measure effectiveness and evasion resistance.
• Failure analysis & deployment: convert failures into reproducible experiments and collaborate with teams to ship robust solutions at scale.

Skills

ML foundations

Python & PyTorch

Agent RL & safety

Experiment pipelines

Hypothesis testing & reporting

Education

BS in CS/EE/Math/Physics or related

MS in CS/EE/Math/Physics preferred

Job description

ByteDance in San Jose is seeking a Research Scientist Graduate (Seed Model‑Alignment & Preparedness) for a 2027 start. You will join the Technology team as a regular, full‑time hire to push forward alignment, safety evaluation, and preparedness for frontier models and agent-enabled systems.

As a new graduate, you will tackle frontier challenges, design agent RL experiments, and contribute to automated red‑teaming, monitoring, and failure analysis.