Senior AI Engineer — Remote Agentic AI for Fintech

Pennylane SAS; France; À distance; EUR 90 000 - 140 000; Plein temps

Avantages offerts par ce poste

25 vacation days

Company shares

Home office budget

Coworking allowance

Gymlib access

Busuu for language learning

Apple equipment

Résumé du poste

Pennylane is seeking an AI Engineer to join our ML & AI teams, building agentic systems and scalable LLM-powered workflows for accountants and SMEs. You will own end-to-end development from intent understanding to tool orchestration and evaluation, collaborating with product and domain experts across Europe.

You will work with Python, MCP, Kubernetes & AWS, focusing on production-grade reliability, memory management, and governance while advancing Pennylane’s AI platform and roadmap.

Qualifications

• 5–8 years of experience and very strong in Python
• Hands-on experience building LLM and agentic systems at scale in production
• Good grasp of applied LLM / ML and AI infrastructure
• Nice to have: model fine-tuning and post-training (SFT, DPO, RL), MCP, familiarity with Ruby
• Fluent in English (French not mandatory)

Responsabilités

• Build and harden agentic loops in production: LLM call orchestration, tool calling, streaming and events, error recovery, checkpointing and resumability, with RBAC, audit trail, and cost controls
• Own context and memory engineering: context construction, retrieval, compaction, memory management and latency/cost trade-offs
• Make Pennylane usable by agents by turning business capabilities into documented, versioned, tested tools exposed internally and through MCP
• Treat evaluation as a first‑class discipline: create datasets, run human eval, track regressions and improve production failures
• Collaborate with Product teams and accountants on live use cases such as ComptAssistant, Studio Assistant, MCP, document extraction and Autopilot bookkeeping

Description du poste

Pennylane is seeking an AI Engineer to join our ML & AI teams, building agentic systems and scalable LLM-powered workflows for accountants and SMEs. You will own end-to-end development from intent understanding to tool orchestration and evaluation, collaborating with product and domain experts across Europe.

You will work with Python, MCP, Kubernetes & AWS, focusing on production-grade reliability, memory management, and governance while advancing Pennylane’s AI platform and roadmap.