Protege is building a platform to solve the unmet need in AI for accessing training data. The Product Manager will own the delivery mechanisms for customers to access and receive data securely and efficiently, while also shaping the transition to a product-led platform for self-service access.
Responsibilities:
- Define delivery as a product capability
- Own delivery surfaces: full contractual deliveries, samples/evaluation datasets, secure file transfer or controlled access environments, and API/structured delivery pathways as appropriate
- Establish what “delivery” means in-product (not just an operational handoff)
- Partner with the Privacy, Rights & Trust PM to embed eligibility checks and usage constraints into delivery workflows
- Ensure access mechanisms enforce policy (without owning policy definition)
- Identify repeated friction surfaced by Solutions Architecture and standardize packaging, formatting, and transformation patterns. Shift delivery from manual, deal-by-deal construction to configuration-driven workflows
- Define what can be self-serve, for which customers, under what guardrails, and with what controls/visibility
- Start with constrained, high-confidence surfaces (e.g., samples) and expand intentionally
- Define delivery status models and ensure internal visibility into what was delivered, to whom, under what constraints
- Enable auditability and reproducibility
- You’ll work closely with: PM, Data Inventory & Partner Systems: ensure what’s discoverable is deliverable, PM, Privacy, Rights & Trust: enforce eligibility and compliance at access time, GTM & Partnerships: improve sampling and evaluation velocity, Engineering: owns ingestion execution and infrastructure
Requirements:
- PM experience in data platforms, marketplaces, APIs, or enterprise delivery systems
- Experience building secure access workflows or permissioned systems
- Strong product + execution instincts; comfortable with operational reality
- Comfortable at the intersection of legal, engineering, and GTM
- Ability to stage maturity over time rather than overdesign upfront