Arseniy Kirykouski · AI engineer, Milan
I build AI systems that see, read and act, then ship them to production.
From researchto runtime.
I’m Arseniy, a final-year AI student in Milan and Member of Technical Staff at Veliu, where we build the agentic commerce layer. I work the whole path: models, evals, APIs and the interface people touch. Hackathons are where I test ideas against a deadline; a few ended on the podium.
My journeyFeatured
Work
Selected builds, from hackathon weekends to production. Models, agents and the products around them.
→Live Context Collision2026
Real-time meeting agent that surfaces conflicting organizational memory while a decision is still being made: a prior legal block, a capacity conflict, a decision the company already reversed.
→Clario2026
Local-first symptom diary. A fine-tuned Gemma model reads the patient’s own timeline (notes, labs, photos) and flags patterns that are easy to miss in a 15-minute appointment.
→Reviva2024–25
Post-stroke rehabilitation from a smartphone camera. OpenCV movement recognition plus a custom PyTorch model that gives patients real-time exercise feedback from visual data alone.
→SMP2DIAG2025
Clinical NLP classifier predicting 22 medical diagnoses from patient symptom descriptions (GretelAI dataset, 1,065 samples). Fine-tuned transformer models: BERT/ClinicalBERT (97% acc, F1=0.97) and Flan-T5 (94% on validation), outperforming classical baselines.
→Bike Sharing Demand Prediction2025
Regression models to predict hourly London bike sharing demand from weather, time and calendar features. Compared linear regression, ElasticNet, Random Forest and XGBoost with nested cross-validation; a tuned Random Forest reached RMSLE 0.30 on the test set (a log-scale error; lower is better).
Five years of
turning questions
into systems.
01 / 05 · 2021 · Tbilisi
International Baccalaureate
42/45, top 3%. Maths and physics focus.
02 / 05 · 2023 · Milan
BSc Artificial Intelligence
Joint degree across Pavia, Milano Statale and Milano-Bicocca. GPA 27/30, ML & DL 27, statistics 28.
03 / 05 · 2024 · Reviva
Team lead, computer vision
Stroke-rehab movement recognition from a phone camera, trained on visual input alone. Winner, Sopra Steria International Student Challenge. Separately, 3rd of 250 at EIT Health i-Days with an LLM MVP.
04 / 05 · 2025 · Madrid · Bologna
Podiums and a 97% model
3rd of 850 at EIT Health i-Days Madrid with Reviva. Won the Biomedicine & AI track at Bologna StartUpDays. BERT/ClinicalBERT at 97% accuracy (F1 0.97) on 22 diagnoses.
05 / 05 · 2026 · Veliu
Member of Technical Staff
Production LLM and agent systems for agentic commerce. At Milan AI Week 2026: 2nd in two Vultr categories and a Google DeepMind special-track prize.
Next chapter
Your team?
Experience
Apr 2026 – Now · Milan, Italy
Veliu
Member of Technical Staff
Veliu is an agentic commerce lab: it makes brand catalogs readable and buyable by AI shopping agents.
- Build production LLM and agent systems, from prototype to deploy
- Design the evaluation and tooling used internally to inspect, compare and improve live product surfaces
- Ship TypeScript and Python across data pipelines, APIs and CI
2024 – 2025
Reviva
Team Lead
Co-led the ML and computer-vision pipeline for an AI-powered post-stroke rehabilitation app: patient movement recognition from smartphone camera data, with real-time exercise feedback from visual input alone.
Recognition
- 2026Milan AI Week hackathon (AI Olympics)2nd in two Vultr categories and a Google DeepMind special-track prize with Live Context Collision, a real-time meeting agent2nd ×2 + DeepMind prizeMilan
- 2025EIT Health i-Days2025 edition, Reviva: core idea and functional demo3rd of 850Madrid
- 2025Bologna StartUpDays, Biomedicine & AI trackWorking real-time prototypeTrack winnerBologna
- 2024–25Sopra Steria International Student ChallengeReviva, post-stroke rehabilitation appWinner
- 2024EIT Health i-Days2024 edition: MVP with an LLM API3rd of 250Milan & Budapest
Toolbox
Programming
ML / Data
Techniques
Tools
Education
2023 – Present
B.Sc. Artificial Intelligence
University of Pavia · Univ. of Milan · Univ. of Milano-Bicocca
GPA 27/30
2021 – 2023
International Baccalaureate
New School, International School of Georgia
42/45, top 3%
Languages
English
C1
Italian
B1
Russian
Native
Belarusian
Native
Got a hard AI problem? Send it over.
Let’s buildtogether