Pavel Vasilyev
LLM Engineer / AI Infrastructure Specialist
Summary
LLM engineer with 14 years ML/NLP experience and 3 years specializing in production language model systems.
Expert in LLM workflows, safety guardrails, prompt engineering, and scaling generative AI infrastructure.
Strong functional programming background (Clojure, Elixir) with production systems experience in Python, Rust, Go, Java, and R.
MSc Computational Linguistics, BSc Nuclear Physics.
Experience
Independent Consultant
Various Clients | Mar 2024 – Dec 2024
- **LevEhat NGO (CTO, 5 months):** Led cloud migration from GCP to AWS, coordinated team of 5 for volunteer management platform
- **Fintilligence (Technical Lead, 4 months):** Built market microstructure analytics platform with PIN/VPIN algorithms and Level 2 data pipelines
- **General consulting:** Time-series forecasting models, NLP classification systems, infrastructure architecture
AI Architect
Stamina AI | Jan 2023 – Dec 2023
- Built complete LLM pipeline: prompt engineering, context management, response generation, safety guardrails
- Integrated OpenAI GPT-3.5/4 APIs with custom safety layers and content filtering
- Designed conversation state management and session handling for therapeutic context
- Set up production infrastructure: API gateway, load balancing, monitoring, logging
- Coordinated with consulting psychotherapists to ensure clinical appropriateness of responses
- Implemented usage analytics and conversation quality monitoring dashboards
Independent Consultant
Various Clients | 2022 – 2023
- Projects: time-series forecasting, text classification, data pipeline optimization
Senior Researcher
Spring Research | 2020 – 2021
- Developed ML models for trading signal generation using time-series analysis and statistical methods
- Built data pipelines processing Level 2 market data (tick-by-tick, order book, market depth)
- Researched topology-inspired approaches to market microstructure modeling
- Implemented backtesting infrastructure for strategy evaluation
Data Scientist
Nestlogic | 2019
- Built computer vision models for advertising creative optimization (image feature extraction)
- Implemented A/B testing infrastructure using statistical hypothesis testing (t-tests, chi-square)
- Deployed ML models to production on Google Cloud Platform with Kubernetes
- Developed analytics dashboards tracking model performance and business KPIs
Data Scientist
Maverick Medical AI | 2018
- Developed NLP system for medical named entity recognition in clinical text using spaCy and BiLSTM
- Built medical ontology framework for standardizing terminology across different hospital systems
- Created decision support tools for clinical workflows highlighting critical findings
- Worked within HIPAA compliance requirements for healthcare data