FERMÀT is looking for a Software Engineer to build out Pierre, an AI Agent, into a production-grade intelligence engine to power personalization, automation, and operational insights across their commerce platform.
Requirements
- 12–18 months building and shipping LLM-powered products (from prototype to GA) with meaningful customer adoption.
- Strong Python and TypeScript experience.
- Solid systems design and API fundamentals.
- Hands-on with LLM APIs, embeddings, vector databases, retrieval patterns and agent/tool design.
- Pragmatic prompt engineering and prompt/version control
- Production chops: cloud (GCP), containers, CI/CD, monitoring/alerting; comfort owning services in prod.
- Experience with streaming responses, Streamable HTTP/SSE/WebSocket patterns for real-time AI interactions.
Responsibilities
- Take AI products to market: prototype quickly, instrument, A/B test, and iterate based on quality, conversion impact, latency, and cost.
- Own LLM orchestration: multi-agent systems, tool use, function-calling, structured outputs, routing, context windows, retrieval and evals.
- Build evaluation & observability: golden datasets, automated evals, prompt/version management and monitoring.
- Productionize reliably: scalable APIs/services, workers/queues, caching and provider failover. Optimize performance while keeping costs reasonable.
- Data foundations: integrate first-party events and catalog data; respect data contracts.
- Vendor strategy: evaluate and integrate multiple model providers (e.g., OpenAI, Anthropic, Google) while tracking quality, cost, and reliability.
- Raise the bar: document best practices, lead design reviews, and mentor engineers on AI patterns.
Other
- Product-minded Software Engineer
- Product sense, clear communication, and a bias to ship.
- E-commerce, recommendations, ranking or ads experience.
- Front-end (React) skills for rapid prototyping of AI UX.
- Experimentation/analytics tools (e.g., PostHog or similar).
