Staff AI Engineer - Agentic Shopping at Picnic

on-site · full-time · Visa sponsorship

Apply for this role at Picnic

Responsibilities:
- Own technical direction and architecture for agentic shopping: design multi-turn conversational systems that ground LLM outputs in product, pricing and operations.
- Build and productionize components: prototypes → stable systems for retrieval, ranking, fine-tuning, context management and deterministic fallbacks.
- Define and implement evaluation, observability and safety loops measuring mission completion, basket correctness, clarification quality, latency and cost.
- Solve hardest production challenges hands-on and mentor engineers across teams on model strategy, inference trade-offs and robustness patterns.
- Collaborate with product, design and ops to align agent capabilities, user experience and operational constraints.

Requirements:
- Proven track record designing and shipping multi-turn conversational/agentic systems to production at scale.
- Hands-on experience training/fine-tuning open-weight LLMs and building RAG/context-management systems.
- Strong engineering skills in Python, PyTorch and production ML tooling; experience with retrieval stores and inference stacks.
- Experience creating continuous evaluation/monitoring and automated safety/guardrail mechanisms for customer-facing agents.
- Demonstrated technical leadership: setting model strategy, performing design reviews and mentoring without relying on hierarchy.