Senior ML Engineer

About the role

Join the AI team at Cosuno and build intelligent systems that power the core of our construction procurement platform. You will take models from concept through production, collaborating across product, engineering and data teams.

Key responsibilities

• Design, train, evaluate, deploy and iterate AI models for construction, price prediction, ranking and search.
• Develop and improve LLM‑based agents that handle multi‑step reasoning in tender workflows.
• Ship models as backend services, APIs and integrate them into our TypeScript web application.
• Collaborate with product, engineering and data teams to translate business problems into ML features.
• Maintain, monitor and continuously improve deployed ML/AI systems.
• Run offline evaluations and online A/B tests to drive metric‑led iteration.

Required profile

• 5+ years building, training and shipping production ML models (classical and deep learning) using PyTorch or similar.
• Strong Python programming experience with production‑grade code.
• Hands‑on experience developing and deploying LLM/agent systems, including RAG and tool use.
• Proven ability to design and run offline metrics and online A/B tests.
• Experience delivering end‑to‑end features from research code to production services and front‑end integration.
• Comfort with ambiguity and deciding when a solution is ready to ship.

Required skills

• Python
• PyTorch
• TypeScript
• React
• Node.js
• Large Language Model (LLM) development
• Retrieval‑augmented generation (RAG)
• Offline evaluation metrics
• Online A/B testing
• Search/ranking/recommendation systems
• Pricing model development
• Embedding models
• Model fine‑tuning
• Scalable model training

What we offer

• Shape a product that solves real‑life problems in the construction industry.
• Open‑minded, collaborative, dynamic, international team.
• Fully remote work culture with flexible hours.
• Company off‑sites and regular team events (offline and online).
• Competitive compensation, permanent contract and modern equipment.
• Open feedback culture with regular 1‑on‑1s.