Dmytro O.

Senior Software Engineer | AI & Data Architect

I build AI and data systems that actually make it to production. Over the past 12 years I've helped startups and scale-ups across the US, UK, Israel and Europe turn complex data and AI challenges into working systems. Currently independent. Getting an LLM to produce something impressive is easy. Getting it to produce the same correct answer every time, on a live surface, where a wrong answer has real consequences - that's the engineering. Most of my recent work is building that layer: the harness around the model. My focus areas: - AI-powered applications (LLM, RAG, vector search, agents and workflows) - AI harness design (guardrails, evaluation, structured outputs, prompt injection, human-in-the-loop) - Data platforms and lakehouses (Databricks, Snowflake, BigQuery) - High-performance APIs and backends - Cloud infrastructure Recent projects: - Conversational AI assistant for a UK recruitment group. Prospective clients describe a hiring scenario in plain English and get an indicative engagement recommendation back. - Built AI-powered search API with hybrid vector+semantic ranking (1M+ CV files across 14 brands; UK recruitment platform) - Built a custom EOR (Employer of Record) classification rules engine with LLM-generated rationales, and a conversational AI assistant in front of it that turns a plain-English scenario into a recommendation while the deterministic engine stays the source of truth (UK recruitment platform). - Architected conversational analytics platform on Azure with medallion data lakehouse (US logistics) - Re-engineered 100+ ETL pipelines, cutting infrastructure costs by 10x (Netherlands energy sector) - Created data platforms for FinTech clients using Snowflake, Databricks, and modern BI tools I combine deep technical expertise with a product management background (DraftKings, Hypatos, Sift). I don't just build systems, I make sure they solve real business problems and deliver measurable ROI. Let's talk if you need: - Production AI/LLM systems that actually ship - Scalable data platforms and pipelines - A technical partner who knows how to take a product from 0 to 1