Machine Learning Engineer (Agentic AI & Automation)

Stealth AI Startup · Milano, Lombardia, Italia ·


Descrizione dell'offerta

Machine Learning Engineer (Agentic AI & Automation)


Location: Italy (Hybrid / Remote within Milan area)

Employment: Full-time · Permanent

Recruitment handled by: Independent search consultant on behalf of a confidential early-stage client


About the Company


Our client is a fast-growing early-stage technology company developing an AI-first platform for operational automation .

Their product combines software engineering, data, and applied AI to streamline processes in a complex, data-intensive sector.


They are now expanding their team with a Machine Learning Engineer who will take ownership of designing, evaluating, and scaling agentic AI workflows built on modern frameworks.


The Role


The company is developing AI agents that automate decision-making and process execution.

You will work on designing, implementing, and fine-tuning these agents, analysing how they perform in production, and improving their behaviour through evaluation and iteration.


This position requires both machine learning understanding and software engineering discipline .

You’ll be joining a small, highly technical team where initiative, autonomy, and experimentation are expected.


Key Responsibilities


  • Build and maintain AI agents and automation workflows using the Mastra framework (TypeScript).
  • Design evaluation pipelines and metrics to track model performance and behaviour.
  • Conduct fine-tuning, prompt iteration, and workflow optimisation to improve outcomes.
  • Integrate LLMs, vector databases, and external APIs into structured agentic workflows.
  • Develop new automation use cases in collaboration with the engineering and product teams.
  • Document experiments and results to ensure reproducibility and transparency.


Tech Environment


  • Core: TypeScript, Mastra (agentic AI framework)
  • Supporting tools: LangGraph / LangChain (Python), vector DBs such as LanceDB or Pinecone
  • MLOps concepts: evaluation, scoring, telemetry, CI/CD, prompt versioning
  • Infrastructure: Node.js, GCP, Docker, Cloud Run, Pub/Sub, Dataflow


Knowledge of Mastra is not required, experience with any agentic or LLM-based framework is acceptable.


Candidate Profile


  • 2–6 years of experience in machine learning, applied AI, or software engineering with ML components.
  • Understanding of LLMs, embeddings, RAG, and agentic workflow concepts .
  • Strong coding skills (Python or TypeScript).
  • Practical mindset and willingness to experiment, test, and iterate.
  • Capable of working autonomously in a non-structured, startup environment .
  • Comfortable analysing data, debugging models, and communicating results clearly.


Desirable Skills


  • Experience with workflow orchestration , LLM evaluation , or MLOps .
  • Familiarity with document automation or structured data extraction .
  • Prior exposure to startups or small engineering teams.


Ideal Candidate Mindset


We’re looking for a builder, someone who enjoys solving open-ended technical problems, exploring new frameworks, and turning research into working production systems.

If you like fast feedback loops, autonomy, and the idea of being the first engineer in a growing AI vertical, this environment will suit you well.

Candidatura e Ritorno (in fondo)