Founding AI Engineer

Mondrio is a fast-growing Seed stage startup focused on building an AI-native pricing management platform. The Founding AI Engineer will be responsible for developing evaluation systems and feedback loops for AI pricing recommendations, ensuring the AI's quality is measurable in customer revenue.

Responsibilities:

• Build evaluation for AI pricing recommendations: eval harnesses and benchmarks that use tracked pricing outcomes as ground truth. Expert review is manual today, and you make it systematic
• Take AI personas further. They simulate B2B buying committees and behavioral effects such as new versus existing customers, grounded in usage data and call transcripts. Automate the parts of persona training that are still manual
• Own LLM infrastructure: routing across current (Anthropic and Google) and future models, with explicit cost, latency, and quality tradeoffs
• Maintain infra and data residency boundaries (e.g. model calls for EU customers must remain within the EU) as we add providers and scale up operations
• Extend the MCP server that LLM agents, including our customers' own agents, use to drive the platform. A feature is done when an agent can drive it through MCP, not when the React component renders
• Work within our typed ontology of pricing entities (Pydantic models for SKU, Proposition, Persona, Quote) so model outputs land in structured, auditable form
• Pinpoint systemic latency, data drift, or cold-start issues in the continuous pricing loop
• Baseline current prompt and model outputs against our typed ontology to prepare for release-gating evals
• Conduct code reviews and lead a technical session on advanced AI/ML patterns for the team
• An eval harness runs on every model or prompt change, and the team trusts its benchmarks enough to gate a release on them
• Manual steps in persona training now run as an automated pipeline built on the same usage data and call transcripts
• Tracked pricing outcomes feed back into recommendation quality, so evals measure revenue impact, not proxy scores
• Participate actively in interview loops to scale the engineering team and mentor mid-level engineers