Work

• Causaly: Staff ML Engineer. I build and run a range of ML and AI-agent systems for biomedical research and analysis - domain where a wrong answer isn't an acceptable norm but a liability.
• Meta: Senior ML Engineer, two teams: (1) Abusive Accounts Detection - built the team's framework for online model training and automatic deployment; owned the team's ML direction and acted as the org's ML-infra point of contact. Domain that is actively adversarial to the deployed models - so they must adapt unattended. (2) Language Understanding for Relevance - short-form video genre prediction, LLMs for entity linking and extraction.
• Yandex: app recommendation algorithms and backend for Yandex.Launcher. Contributed to the open-source release of Catboost, Yandex's gradient boosting library.
• Segmento: click-prediction models for online advertising and evaluation methodology for them.
• Aintsys: R&D on neural networks for financial time-series prediction and trading strategies.
• EFA Medica: firmware for electrosurgical devices. Designed a universal TFT screen module that replaced specialised LED indication and control modules across the company's entire next generation of devices. Domain where software absolutely cannot fail: it controls surgical scalpel.