About the role
Senior AI Engineer at Pennylane, one of the fastest-growing French fintechs, building the financial operating system for European SMEs and accounting firms (1M+ SMEs, 6,000+ accounting firms, 400M euros raised including from Sequoia).
You would join one of the ML and AI teams (25+ people, growing) working on the agentic layer of the product: the agent harness and runtime, model gateway, tool registry and MCP, guardrails, observability and the Studio Assistant orchestrator. The core of the job is building and hardening agentic loops in production: LLM orchestration, tool calling, streaming, error recovery, checkpointing, plus context and memory engineering with the latency and cost trade-offs that come with them. Evaluation is a first-class discipline (golden datasets, LLM-as-judge, regression tracking). The stack is Python, agentic frameworks and MCP, evaluation and observability tooling, on Kubernetes and AWS.
Highlights
- Agentic layer: agent harness and runtime, model gateway, tool registry and MCP, guardrails, observability, Studio Assistant orchestrator
- Python, agentic frameworks and MCP, evaluation and observability tooling, on Kubernetes and AWS
- 1M+ SMEs, 6,000+ accounting firms, 400M euros raised including from Sequoia
- Fully remote within Europe
This recap is dataskew's editorial summary, not the company's copy.