Software Engineer with ML and Data Skills

VirtusLab is a leading European software consulting and engineering company. Our mission is to craft clean code and practical solutions with precision and purpose. We foster a dynamic culture rooted in strong engineering, a sense of ownership, and transparency, empowering professionals to make a substantial impact in the software industry.

About the role

Productionizing and scaling an ML-driven data quality system across the organization. The scope of services involves: building and tuning anomaly-detection and clustering pipelines, pairing classic ML with LLM reasoning to flag and explain issues, collaborating with data producers to fix root causes, and creating as well as maintaining validator models that turn detected anomalies into better future data.

Project Scope

Our client is a NASDAQ-listed B2B data company powering Go-To-Market strategies with a 360-degree view of every customer, a view whose value depends on the quality of billions of person and company records.

Anomalsky is the ML system built to catch what traditional observability misses: row-level semantic anomalies (e.g., a first_name, title, company_name). Three layers, an ML layer (embeddings + unsupervised clustering) flags suspicious records at scale, an LLM layer removes false positives and explains each cluster, and an optional human-in-the-loop lets domain experts resolve whole clusters at once. The MVP already drove ~40k crucial record corrections in production.

What’s next: the MVP is landing on GCP now. Once it’s operational, the mission is to scale Anomalsky across the entire organization, embedding it into Acquisition pipelines and building a real-time variant that scans data before it reaches customers.

The scope of cooperation covers

• Productionizing Anomalsky on GCP and scaling it to operational, organization-wide use.
• Evolving the ML / LLM / human-in-the-loop design and the feedback loop that turns expert reviews into reusable knowledge.
• Prototyping the low-latency real-time variant.
• Integrating Anomalsky into existing workflows, starting with Acquisition.

Tech Stack

Python, Airflow, BigQuery, Snowflake, Spark (Dataproc), Databricks, Iceberg, Starburst, Trino, AWS, GCP, Docker, Terraform, Jenkins, GitHub, Scikit Learn, unsupervised anomaly detection (kNN, Isolation Forest, autoencoders), recursive clustering, classifiers on real + synthetic data, MLflow, LLM-based reasoning.

Project environment

ML and data engineers from VirtusLab collaborating with customer data engineers and product management.

A few perks of being with us

• Building tech community
• Flexible hybrid work model
• Home office reimbursement
• Language lessons
• MyBenefit points
• Private healthcare
• Training Package
• Virtusity / in-house training
• Access to the above perks is optional and completely voluntary for B2B contractors