ML/AI Engineer who takes systems from research to production. LLMs, computer vision, large-scale search — built and shipped across e-commerce, defense tech, and fintech.
IMO medalist. Kaggle Master. 10+ years building AI/ML systems at scale. I solve hard problems because I trained on hard problems.
You bring the problem — I bring the solution. I analyze, architect, and implement end-to-end. Then I hand over a working system, not a deck.
Technically challenging ML over massive data. I've built AI-powered search over petabytes of photos and classified 100B posts for under $3K. I've replaced 300 production models with one, saving $100K/year. If the problem has scale and needs precision engineering — that's where I operate.
Building an internal coding co-pilot as a multi-agent system for automatic code generation in a custom DSL. Uses LLM-as-a-Judge, agent collaboration, dynamic context auto-compaction, and custom orchestration.
End-to-end AI profiling pipeline that aggregates massive volumes of posts, reposts, likes, and comments across social platforms, filters signal from noise, and synthesizes balanced intelligence dossiers with precise detail — flagging suspicious patterns while avoiding false positives.
Compressed 300+ individual classification models into a single multi-taxonomy multitask architecture using a shared architecture — cutting cloud costs by ~$100K/year while improving accuracy through knowledge transfer across taxonomies.
Designed and built visual retrieval across ~3 billion public photos. Identity search and entity resolution for intelligence consumers — from a single photo to matching appearances across social platforms.
AI-based search across 100 billion posts and comments, flagging suspicious pro-Russian and extremist activity. Detects threats from individuals and organizations — including anticorruption use cases.
Built forecasting systems serving predictions for a 10M+ consumer panel, alongside ML platform infrastructure on AWS for model serving and training.
Fully automatic book-to-visual pipeline. Smart genre detection, character and scene extraction, and AI-generated stunning visuals — delivered as Instagram-style stories or cinematic video trailers.
Multi-agent code generation, multitask classification (300+ → 1 model), LLM systems
AI search pipelines over 100B posts, petabyte-scale face search, AI profiling for OSINT
Forecasting at 10M+ scale, ML platform on AWS/K8s
Social media monitoring, NLP models (sentiment, NER)
LiDAR / 3D point-cloud semantic segmentation
Credit scoring, fraud detection, NLP R&D
Network monitoring systems, low-level algorithms
Architect-level or senior ML/AI engineer position. Part-time, fractional, or contract arrangements welcome.
Based in Kyiv, Ukraine. Remote-flexible — experienced working across time zones with distributed teams.
AI, Agents, LLMs, efficient ML solutions, scale. Comfortable anywhere from R&D to production deployment.
10+ years shipping AI/ML systems. Research-to-production delivery. End-to-end ownership from modeling to deployment.
Fully automatic book-to-visual pipeline — genre detection, scene extraction, AI-generated visuals as stories or cinematic trailers.
AI shopping assistant for bookstores — guides customers through a short conversation and recommends the perfect book from the catalog.
Multi-agent AI for human-like LinkedIn content — voice personas, creativity randomization, and a judge that rejects anything that reads like AI.
Thoughts on ML engineering, building production AI systems, and lessons from shipping at scale. Stay tuned.