Software engineer and neural-network specialist with 5+ years building AI into B2C and B2B products and leading delivery on high-impact open source. I design and own systems end to end — backend services and APIs, desktop applications, and the models behind them — from architecture to deployment. My focus is generative and language models, applied ML and scalable backends in Python, React Native and JavaScript. Author of 20+ scientific publications, 9 registered programs patents, and research indexed in Scopus and Web of Science.
Neural networks integrated in a single product
Scientific publications
Software patents in FIPS
Publications in Scopus & WoS
One engineer, the full stack of an AI product
I take products from idea to production — and I’m just as comfortable in a research notebook as on a deployment server.
Backend & Architecture
Scalable services and clean API design in Python, system architecture, traffic analysis, DevOps and server administration, VPN/networking and deployment.
Apps & Interfaces
Desktop applications, web, PWA and mobile apps with React Native, rich JavaScript interfaces, Telegram bots, AR/VR and games with neural agents.
Neural Networks & ML
Designing architectures and training generative, LLM and computer-vision models — including on-prem agentic “thinking” AI — from data pipelines to real-time inference in production.
Languages
AI / Machine Learning
Backend & APIs
Apps
Infrastructure
Leadership
Experience building products for real users
-
View on GitHubWunjo Make
A neural-network suite with thousands of active users that integrates 50+ models — face swap, lip sync, voice cloning, object/text/background removal, restyling, audio separation and video generation — across desktop and web. MIT-licensed, top weekly and monthly product rankings.
-
View on GitHubMOEX Portfolio
A web app that analyses Moscow Exchange stocks with neural networks and technical indicators to build and track an investment portfolio — data ingestion, model training and a web dashboard in one pipeline.
-
Get it on Google PlayDoughfolio
A published mobile app for building and tracking a crypto investment portfolio — shipped end to end on Google Play, from data and logic to a clean mobile interface.
-
Open the productWunjo Design
A web product that generates editable vector graphics with neural networks — an innovative take on AI design that outputs clean vectors rather than raster images. Built the full stack: backend, generation pipeline and the browser editor.
More open source: Berkano (plant-disease ML app), a Voice Assistant for Raspberry Pi / Jetson, and Neural Radio.
Where I’ve built things
-
Aug 2022 — Present, Saint Petersburg
Senior Software Developer — VAS Experts
Build fundamentally new neural-network solutions and ship them to production: an on-prem “thinking” search assistant over internal documentation and closed tasks, powered by a local LLM, with automated server diagnostics; and a deep-learning system that analyses network traffic to detect VPN signatures for blocking or policing — the basis of the third Scopus paper. Work spans LLMs, CNNs and computer vision.
-
May 2024 — Present, Individual Entrepreneur
Founder & Open-Source Maintainer
Build and maintain open-source and commercial AI products with thousands of users — Wunjo Make (1.2k★) and the new Wunjo Design — integrating 50+ neural networks. Own the full lifecycle from architecture to deployment and support, with recognition in top weekly and monthly product rankings.
-
2021 — 2022, Anapa
Software Developer — Freelance
Delivered custom web applications and integrated neural networks for clients — including computer vision for a drone that recognises objects and estimates distance — with model training, data science and team management.
-
Sep 2020 — Jul 2021, Saint Petersburg
Developer RDP — Tricolor TV
Applied machine learning to television-viewing behaviour — classifying subscribers by what they watch to inform which ads to show on TV.
2019 — 2021, Saint Petersburg
Master of Engineering
ITMO University
Diploma: training and servicing space-station systems with Virtual Reality and ML for real-time simulation and predictive analytics.
2018 — 2019, Italy
Master’s — Exchange
Polytechnic University of Bari
Research in non-destructive testing (NDT) with innovative thermography methods not yet applied in industry.
2014 — 2019, Novosibirsk
Bachelor of Engineering
NSTU
Diploma: a novel thermography NDT method for advancing industrial thermal-imaging diagnostics.
Data Science Certificate
ITMO University, 2022
Project Management Certificate
MFUA, 2020
QA Training Certificate contest winner
Yandex Academy, 2019
“The interesting part of AI isn’t the model — it’s making it run for real people, on real hardware.”
How it unfolded
2026 — now
Wunjo Design
The newest product — a from-scratch PWA where neural models generate editable vector designs, not flat images or un-editable HTML.
2024 — now
VAS Experts
An in-house, Gemini-style search over docs with AI Agents and tickets on a local LLM. I planned the architecture, procured hardware, wrote the API, cut LLM load, fine-tuned LoRA on our docs and trained from-scratch models that imitate our engineers — it reasons and asks clarifying questions, and I extended it to diagnose DPI servers from traffic logs.
2023 — now
Wunjo Make
A neural-network suite that grew to a lot models as voice cloning, lip-sync, face swap, restyling, video generation. Open source (over 1.2k★) with a monetized commercial version (5k+ users).
Open source
Berkano
A plant-analysis network trained in Python, then compressed from a 2 GB model to run via React Native (ONNX Runtime) on 2023-era phones — not just the newest ones.
Show earlier (2018 — 2023)
Spin-off
Drone swarm
On an experimental swarm project (successful enough to spin out a company) I built the drone’s brain on NVIDIA Jetson like object & face recognition with depth and distance estimation, fusing models to stay real-time without overloading the hardware.
2022 — 2024
VAS Experts
A new convolutional-network algorithm that classifies VPN vs. clean traffic from IPFIX behaviour.
Open source
Neural Radio & MOEX Portfolio
Infinite generated lo-fi radio (up to ~500 simultaneous listeners), written in Go with models trained in Python on a separate server and MOEX Portfolio over market data and comment sentiment to score a security’s rise/fall (it worked, but the edge was thin — an honest lesson in real-world ML).
2021 — 2022
Custom development
Camera systems that recognise faces and detect people’s presence and movement.
2020 — 2021
Tricolor TV
Imputing answers viewers chose not to share and scaling household behaviour to the whole audience, to target ads by channel and time slot.
2019 — 2021
ITMO University
A dedicated ML and AI track in Python. My thesis adapted model-evaluation methods to score how well people learn action sequences in Virtual Reality.
2018 — 2019
Politecnico di Bari (Erasmus)
Machine Learning in R, and reinforcement-learning game agents for AR/VR.
Where the work got noticed
1.2k★ on GitHub
The developers appreciated my open source contributions and repositories with stars, reposts, and donations.
5× hackathon wins
Led a team to five consecutive wins from 2020 to 2022 years and expert at the World Festival of Youth 2024 hackathon.
20K blog readers
TenChat award for neural-network content (20k readers) in 2023 and top author on Habr about neural networks.
60+ startups helped
I regularly support early stage startups as a mentor on Product Radar and help them achieve various goals.
Overcoming challenges in deep inspection of VPN and proxy by deep learning
Journal of Physics: Conference Series, 2024, DOI 10.1088/1742-6596/2701/1/012106
The use of virtual reality in the simulator of maintenance and repair of technical components
ASEDU-II 2021, 2022, DOI 10.1063/5.0104094
Research on the effect of test parameters on signal contrast and accuracy of determining the size of a defect
IOP Conf. Series: Materials Science and Engineering, 2021, DOI 10.1088/1757-899x/1019/1/012056
What colleagues say
Vladislav delivered neural-network solutions we’d considered out of reach — traffic analysis, computer vision and language models — and took them all the way to production. Dependable, hard-working and genuinely strong technically.
Alexey Alexeenko
CEO, VAS Experts
One of the clearest voices on applied AI in our community. Vladislav explains complex models simply and backs every idea with working code — a natural communicator and builder.
Oleg Ratunin
CEO, TenChat
A rare engineer who combines real research intelligence with the discipline to ship. Vladislav asks the right questions and turns them into systems that actually work.
Dr. Alexey Fedorov
Professor of RAAI, ITMO University
Curious, rigorous and self-driven. In our lab he introduced methods that were new to us and saw them through to results — a pleasure to supervise.
Dr. Michele Fiorentino
Professor of DMMM, Polytechnic University of Bari
Talented and tireless. His diploma work brought approaches that hadn’t been applied in production before, executed with unusual maturity for his level.
Dr. Sergey Salenko
Professor of FLA, NSTU
Written recommendation letters and contact details are available on request.
Have a role in mind?
I’m open to senior engineering and AI roles. Reach out and I’ll reply with my full CV and answers to anything you need.
