The Role:

We're building Subtrace: an AI-native layer that automatically captures tasks activity and reconstructs it as structured process data: tasks, workflows, bottlenecks, and automation candidates. No integrations. No manual modeling. Just a the data layer that understands work.

What we are looking for:

• 3–5+ years building ML/AI systems in production

• Strong Python skills (data pipelines, model development, APIs)

• Experience with LLMs, embeddings, or sequence modeling

• Solid understanding of data engineering (ETL, streaming, feature pipelines)

• Familiar with cloud environments (Azure preferred) and containerization (Docker/Kubernetes)

• Able to work independently and ship end-to-end

• Strong communication (technical + non-technical)

Nice-to-have:

• Experience with process mining / workflow intelligence

• Background in distributed systems or event-driven architectures

• Experience fine-tuning or deploying LLMs

• Azure ML / MLOps tooling

• Timezone: CET (+/- 3h)

Responsibilities:

• Build and deploy ML models that infer workflows from raw activity data

• Design data pipelines for ingestion, transformation, and feature extraction

• Work with LLMs to structure and enrich unstructured signals

• Own model lifecycle: training, evaluation, deployment, monitoring

• Collaborate closely with product and engineering to ship fast

• Continuously improve system performance, accuracy, and scalability

Why this role:

• Work on a real AI-native product (not a wrapper)

• High ownership and fast iteration cycles

• Direct impact on core product and architecture

• Transparent setup and lean decision-making

• Strong upside via early-stage exposure