5+ years. Zero demos. All production.
Different companies, different stacks.
All of it ended up in front of users.
I handle the model, the backend it plugs into and the MLOps around it, so you don't have to stitch together three different specialists.
Freelance @ STRV
Senior Machine Learning Engineer
Built recommendation engines, content moderation models, and audio classification tools using technologies like Elasticsearch, PyTorch, and AWS Lambda, while delivering talks on MLOps and online learning.
Miya Consulting
Senior Machine Learning Engineer
Built a Slovak telecom's scraping and classification stack: large scale crawling and filtering, custom URL and phone number classifiers, and LLM output distilled into lightweight production models on Azure Synapse and Azure ML.
Stil
Machine Learning Engineer
Created AI systems for tremor classification using IMU data, supported clinical statistics, and managed backend APIs, deployments, and databases under strict cost constraints.
PwC
Data Scientist
Built agentic LLM pipelines for private equity due diligence: key deal issue extraction, engagement matching, RAG over firm economic data, and an MCP server. LLM-as-judge evaluation harnesses halved inference cost on Azure.
O2
Machine Learning Engineer
Developed neural-network and XGBoost models for customer lifecycle segmentation and churn prediction, building large-scale PySpark data pipelines and deploying ML solutions across multiple telecom operators using Azure, Databricks and CI/CD automation.