Petr Polezhaev

- Shipped LLM QA system (web chat, WhatsApp, voice), doubling automation across 10+ markets - Designed and implemented agentic architecture for AI support platform; authored foundational ADRs shaping team roadmap - Led cross-team OTP auth design (5 teams), significantly increasing auto-resolved case volume - Built hierarchical multi-class classifier (+40% accuracy) with custom LLM evaluation framework - Created conversation-level evaluation system driving measurable quality improvements across 10+ markets

- Built LLM-powered email processing (summarization, sentiment, extraction), accelerating response time and support quality - Deployed internal Slack chatbot integrating LLMs with company data for 1,000+ employees - Prototyped contract information extraction system, catalyzing a dedicated project and MVP - Led multiple GenAI/ML PoCs from prototype to stakeholder buy-in, influencing product direction - Designed price steering engine for car rentals using demand forecasting and constrained optimization

– Built core recommender system components: API service, ML models, data pipelines, and monitoring – Created segmentation and offer selection models (multi-armed bandits), increasing client ARPU by up to 23% – Designed A/B testing framework; built Grafana dashboards and alerting for model and business metrics – Introduced code review culture, CI/CD best practices, and comprehensive test coverage to the ML team

1. Research - Built regression models estimating curriculum item difficulty for e-learning students - Developed meta-heuristic algorithms for VM and compute unit scheduling - Designed and built a system for automatic VM disk image creation, including the integration layer with IaaS providers 2. Teaching - Taught courses in machine learning, software development, and computer security

Extended multi-platform CI/CD pipeline with code coverage analysis, memory and threading diagnostics - reducing bugs in the Intel software product.

Diploma of professional development in computer security.

Postgraduate Research – Mathematical Simulation & Software

Qualification: mathematician-programmer, GPA 5.00 (Excellent)