Machine Learning Engineer

Comand AI's mission is to build next-generation C2 software with real users in real deployments in the field. On the ML side, the job is to build decision systems out of unstructured data, mostly documents, with models designed to actually be used in the field.

The scope is end-to-end and the ownership is high: you take a problem from data to a model someone actually uses. Today's work leans LLM and NLP first, with some vision, and spans agent architectures, fine-tuning, open-source models, and constrained deployment (latency, security, on-prem and offline).

What this looks like in practice:

• Build agents with real architecture questions: multi-agent setups, tool management, orchestration
• Solve runtime and scaling problems: background jobs, information flow between systems
• Extract information from documents with little training data and a high bar for quality
• Explore reinforcement learning for maneuver generation
• Help move the team from working sequentially to running several ML efforts in parallel as small squads

Must-have:

• Strong applied ML background with production experience
• Comfortable with LLM/NLP work: fine-tuning, agent orchestration, evaluation
• Solid Python and experience shipping models into real systems
• EU citizenship (contractual requirement tied to our defense contracts)
• Comfortable owning a problem end-to-end, from data to a model in production

Nice to have:

• Computer vision experience
• Experience with constrained deployment: on-prem, offline, or latency-sensitive environments
• Prior work in defense, government, or dual-use tech
• Experience with reinforcement learning

• Competitive package (top 0.1% of compensation in Europe: base + equity)
• Paris-based with occasional travel to active deployment areas