Oleksiy Dolgykh

Oleksiy Dolgykh

About

Detail

AI and ML Expert | Full Stack Data Scientist | Principal Engineer | Engineering Manager and Director
Winnipeg, Manitoba, Canada

Contact Oleksiy regarding: 
work
Full-time jobs

Timeline


work
Job
school
Education

Résumé


Jobs verified_user 0% verified
  • A5 Labs
    Senior AI/ML Engineer
    A5 Labs
    Mar 2026 - Current (8 months)
    Designing AI feedback flywheels and regression/evaluation systems for game products. Building AI-native engineering workflows with MCP tools, Claude/Codex agents, CI/CD checks, and eval-driven automation for safer iteration on LLM behavior. Tech: Python, Pydantic AI, LLMs, FastAPI, AWS, MCP, Eval tools.
  • S
    Head of Artificial Intelligence, AI Engineering Manager
    Synthetic Turing Experience Technologies
    Jan 2025 - Jan 2026 (1 year 1 month)
    Led a team building a GenAI companion platform, branded virtual personas across chat, images, and voice. Built LLM and image generation infrastructure, semantic retrieval, and evaluation that prioritize reliability. Managed teams across AI/ML, backend, frontend, and QA. The challenge was shipping systems that work under real load – architecture that scales and evaluation that catches problems early. Tech: Python, AWS/GCP, FastAPI, PyTorch, vLLM, Pinecone, RunPod/Modal, Flux/ComfyUI, Inference engineering.
  • Toptal
    Senior AI/ML Engineer at Core Team
    Toptal
    Jan 2021 - Oct 2024 (3 years 10 months)
    Architected Toptal's GenAI platform. The problem was that they had lots of AI prototypes but no path to production. Built the infrastructure that changed that, integrating LLMs, RAG, and vector databases. This meant designing serving patterns, implementing retrieval, and creating evaluation frameworks that gave confidence before shipping. A big part of the job was industrializing delivery – re-establishing core operating practices (roadmap, lifecycle thinking, delivery rhythm) – so ML work could move predictably and faster. Tech: Python, GCP, FastAPI/Flask, PyTorch/Tensorflow, HuggingFace, Weaviate/Qdrant, Kubeflow/MLflow/Dagster/Airflow, BigQuery, Vertex AI.
  • Arkadium
    Director of AI/ML
    Arkadium
    Jul 2019 - Dec 2020 (1 year 6 months)
    At Arkadium, my focus was on bringing AI into the business in a way that actually shipped - not as experiments, but as a production product. I took on the responsibility of building a new AI department inside the company and worked closely with stakeholders to turn rough ideas into concrete, production-ready results. One major track was the Family Feud game: I owned the full AI/ML scope and led cross-functional collaboration to get it over the line. I also developed a computer vision algorithm to automate Hidden Object game creation, cutting down the manual effort required from game designers. Tech: Python, Azure, Azure Machine Learning, Kubernetes, flask, Tensorflow, Keras.
  • Arkadium
    Lead Machine Learning Engineer
    Arkadium
    Jan 2018 - Jul 2019 (1 year 7 months)
    I focused on building ML systems that were safe, reliable, and practical to run in production. I cared a lot about turning messy real-world text and data into outputs that the business could trust. A big part of my work was building robust NLP systems at scale. I created a cross-lingual text classification approach across EFIGS languages and cut the error rate by 75%. I also engineered a brand-safety model for sports content that reduced brand risk by 98% - work that was later recognized in the "Media Quality Report". I’m especially proud of the early AI product Inhabit, where I owned the ML end-to-end: turning sports articles into structured understanding (what it’s about, who it mentions, what’s risky) so the product could generate safe,
  • Arkadium
    Director of Engineering
    Arkadium
    Jan 2016 - Dec 2017 (2 years)
    I led a 25-person software and QA engineering team delivering world-famous Windows games like Microsoft Solitaire Collection, Microsoft Mahjong, and Microsoft Minesweeper. A big part of the job was working directly with Microsoft stakeholders - setting clear expectations, driving strong execution, and shipping reliable releases. I also invested heavily in leadership and team health - scorecards, recruiting, mentorship, and helping people grow into leadership roles - while pushing the org toward top crash-free reliability that Microsoft recognized. In parallel, I built out a QA automation function for the department by leveling up a manual QA org into an automation-capable team, which raised quality and made releases more predictable.
  • R
    Data Scientist
    Robotics Laboratory
    Sep 2015 - Aug 2017 (2 years)
    • Researched and developed neural network algorithms for magnetic tracking systems, achieving significant acceleration of coordinate detection time, and enabling real-time application.
    • Co-authored and published several papers on these advancements.
    • Provided consultation to the team on the feasibility and limitations of machine learning techniques within the domain of magnetic tracking.
  • Arkadium
    Engineering Manager
    Arkadium
    Sep 2014 - Dec 2015 (1 year 4 months)
    I focused on making delivery predictable and quality repeatable for high-visibility, consumer-facing games. I introduced stronger engineering habits - code quality practices and continuous integration - so the team could ship with confidence and fewer surprises. I also improved day-to-day execution by moving project work into Azure DevOps, which made collaboration and delivery smoother across the team.
  • Arkadium
    Senior Software Engineer
    Arkadium
    Apr 2012 - Sep 2014 (2 years 6 months)
  • Arkadium
    Software Engineer
    Arkadium
    Sep 2010 - Apr 2012 (1 year 8 months)
  • B
    Software Engineer & Co-Founder
    Babyshop.In.UA
    Oct 2007 - May 2010 (2 years 8 months)
    • Developed one of Ukraine's first custom e-commerce websites, building all features from scratch in pure PHP, including vendor management, warehouse accounting, product catalogs, order management, customer history, etc.
    • Collaborated closely with the business stakeholder (my wife) to design and implement functionalities that weren't available at the time, pioneering unique e-commerce capabilities.
    • Managed end-to-end tech responsibilities, from web development to server deployment and domain management, ensuring a fully operational and scalable online store.
  • B
    Software Engineer
    BOSSdev, Inc.
    Jan 2007 - Aug 2010 (3 years 8 months)
Education verified_user 0% verified
  • RRC Polytech
    Communication for Business and Entrepreneurship
    RRC Polytech
    Jan 2024
  • D
    master, Master's degree, Robotics
    Don State Technical University
    Jan 2015 - Jan 2017 (2 years 1 month)
  • V.I. Vernadsky Taurida National University
    master, Master's degree (WES-evaluated), Computer and Information Sciences, General
    V.I. Vernadsky Taurida National University
    Jan 2000 - Jan 2005 (5 years 1 month)
This is a community-created genome.