Trust & Safety Tech PM: Safe AI Data & RL Environments

Innodata; United States; Remote; USD 180,000 - 220,000; Full time

Job summary

Innodata seeks a Technical Product Manager for Trust & Safety to define and ship data products used to train and evaluate models for personalization and safety tasks. You will own the full lifecycle from lab needs to RL environments, rubrics, evals, and delivery with expert annotators, ensuring models train and perform safely.

You thrive in ambiguity, design trap-based environments with misalignment and cyber traps, and define measurable safety outcomes.

Qualifications

• 5+ years in product management for ML/AI data or safety.
• Hands-on with training/evaluation of LLMs: SFT, RLHF, reward models.
• Familiar with agentic AI and failure modes.
• Strong safety knowledge: reward hacking, jailbreaks, prompt injection.
• Read code and review environment specs; Python and sandboxing a plus.

Responsibilities

• Define data products for safety training data, RL environments and evaluations.
• Design trap-based RL environments with misalignment and cyber traps.
• Specify safe-behavior rubrics, verifiers, and reward signals.
• Define measurable safety metrics and track improvements.
• Collaborate with researchers, ML engineers, red-teamers, policy experts.
• Communicate specs and findings to researchers and executives.

Skills

ML safety PM

Technical PM

Adversarial thinking

Stakeholder comms

Python basics

Tools

Docker

Virtual Machines

Browser automation

Job description

Innodata seeks a Technical Product Manager for Trust & Safety to define and ship data products used to train and evaluate models for personalization and safety tasks. You will own the full lifecycle from lab needs to RL environments, rubrics, evals, and delivery with expert annotators, ensuring models train and perform safely.

You thrive in ambiguity, design trap-based environments with misalignment and cyber traps, and define measurable safety outcomes.