Technical Product Manager, Trust & Safety

The role

You will define and ship data products used by frontier labs to train and evaluate models for personalization, computer-use, and GDPval tasks. You will own the full product lifecycle, including designing reinforcement learning environments with misalignment and cyber traps.

What you will do

• Define data products for frontier labs by turning research goals into product specs.
• Design trap-based RL environments including misalignment and cyber traps.
• Build safety into agentic tasks by specifying rubrics, verifiers, and reward signals.
• Own evals and metrics to define and measure safe and trustworthy products.
• Prioritize product development and technical tradeoffs.
• Partner across teams including researchers, ML engineers, red-teamers, and policy experts.
• Track the threat landscape and turn research into roadmap items.
• Write specs, customer briefs, and safety findings.

Required qualifications

• 5+ years of product management experience, including technical products in ML, AI data, security, or trust & safety.
• Hands-on familiarity with how LLMs are trained and evaluated: SFT, RLHF/RLAIF, reward models, verifiers, and benchmark design.
• Working knowledge of agentic AI: tool use, computer-use agents, multi-step planning, and failure modes.
• A strong grasp of AI safety concepts such as reward hacking, deceptive alignment, jailbreaks, and prompt injection.
• Technical depth to read code, review environment specs, and make technical tradeoffs.
• Experience designing metrics that capture risk, system performance, and user impact.
• A track record of shipping 0-to-1 products in fast-moving, ambiguous environments.
• The ability to explain complex technical ideas clearly to researchers, customers, and executives.
• Creative, adversarial thinking.

Nice to have

• Comfort with Python and sandboxed environments (Docker, VMs, browser automation)

About the company

Innodata is a global data engineering company that provides solutions, platforms, and services for Generative AI builders and adopters.