YEHOR TERESHCHENKO

AI Engineer
Full-Stack, automation and Games Developer | Innovation Enthusiast

Developer with expertise in building production-grade LLM applications, fine-tuning models, implementing RAG and agentic systems. Developer of games and educational programs on Unity.

Work Experience

Co-founder in Osakeyhtiö AI startup company
@ Lemilabs Oy
August 2025 - Present

Developing cybersecurity AI assistant for programmer and DevOps engineers; + other AI projects.

Oracle AI - Transition to another internal project during contract work
@ Metropolia UAS (AI R&D)
May 2025 - Present

Research, Forecasting, Innovation; AI, Python (Flask, FastAPI), Vue.js; Vector DB, Embedding, LLMs, big data, various metrics, trend analysis, and clustering; Multi-agent workflow.

Developing a project that uses different resources (like media, news, jobs), analyze it and forecast the future trends using AI and LLMs techniques.

AI autonomous UAV project
@ AeroRozum founder - toiminimi
April 2025 - Present

Founder AeroRozum. Main project is now in R&D and prototyping stage. Technical PoC complete.

Fully automated drone (UAV) that can fly very long distances with no human intervention or maintenance. Everything, from mission planning & execution to obstacle avoidance, self-charging, and signal transmission, is done by the device itself.

R&D contract work at Metropolia AI project, developer of innovative AI full-stack platform for writing grant applications
@ Metropolia UAS (AI R&D)
January 2025 - May 2025

Vue.js frontend with Python Flask backend, MariaDB SQL, JavaScript, Jinja templating, and LLM integration for automated proposal's text analysis, suggestions, workpackages/risk assesment, keypoints generation generation & validation, etc.
Personally fully developed Google Docs plugin for grant writing. Conducting experiments with LLMs, fine-tuning, RAG, contribution to other projects.

Computer Vision & ML Developer | Project abroad
@ TalTech – Tallinn University of Technology
October 2025

Developed an AI-powered safety detection system for Enefit Industry using computer vision and machine learning. Implemented real-time visual analysis with deep learning models for object detection, anomaly detection, and safety protocol monitoring. Technologies: Python, OpenCV, PyTorch/TensorFlow, CNN models, image processing pipelines, real-time inference. Collaborated in a cross-country team with Enefit Industry representatives and TalTech mentors. Delivered working prototype from concept to demo in a short time. Received "Excellent" rating from company representatives and jury.

Language Learning Game Developer
@ Entrepreneurship, IndieDev
September 2024 - January 2025

Designed a language-learning game integrating book reading, vocabulary building, and engaging monster battles. Used Unity, C#, AI & Translators; Python and DB.

Bot and blockchain data parsing
@ Individual Customer
April 2024 - January 2025

Developed and maintained a production-level commercial TON-parser bot for Telegram. Completed other tasks in blockchain data parsing, AI and related areas using Python, SQL, TON-Chain, and LLMs APIs.

Trading Indicators and Strategies Developer
@ Commercial orders from investors and traders
September 2023 - February 2024

Built Pine Script indicators and Python-based trading bots for automation on platforms like TradingView, Bybit/Binance. Used AI to analyze trading data too. Technologies: Python, PineScript (JS syntax), JS, AI, Trading data, Blockchain.

Physics Lab in the Smartphone Developer
@ R&D in Scientific LIT, IndieDev
May 2023 - June 2024

Created and maintained an educational app for research and development project, enabling physics experiments using smartphones. Used Unity (C#), JavaScript and Python.

Decentralized Public Alert System Developer
@ Entrepreneurship, R&D
June 2022 - July 2023

Developed an alert system leveraging volume speakers for emergency notifications in remote areas in Ukraine. The system allowed people to connect their speakers and receive rewards for notification people depending on place, volume, alert frequency etc. Used JavaScript, Python (Flask&FastAPI), SQL, APIs, Blockchain (Solidity).

Volunteering

Reviewer of articles in scientific journals
2025 - Present

NLP4DH, Journal of data Mining and digital humanities: assessment of scientific papers, decision making on whether to accept the paper into a journal or not, writing a review

Mentor, active member, co-founder at AIvoLink - students club in Metropolia
2025 - Present
President of the Aerospace Center (Lyceum)
2021 - 2024

Official volunteer role awarded by vote and professors recommendation. Led aerospace initiatives, organized events, and coordinated projects to inspire a passion for STEM. Assisted teachers in delivering engaging physics and astronomy lessons. Developed software and created original challenges and tasks. Authored research papers and maintained academic journals.

Cybersecurity and Army Forces of Ukraine
February 2020 - 2022, 2025

All possible help to my country in difficult times. Finding bugs, security holes, access errors, writing software etc. Cooperation with "IT ARMY of Ukraine". Creating a website for automatic DDOS of propaganda resources of the russia, creating protection against attacks, helping people with the configuration of remote servers for that purpose, and assisting in developing military drawings, participating in the production of masking nets.

Security Audit & Vulnerability Remediation for Production AI Tool (mikro-Mikko): I led a comprehensive security audit for a critical production-level tool, identified 5 high-impact critical issues, 4 moderate vulnerabilities, and 2 low-impact issue.

Education

Metropolia University of Applied Sciences
Bachelor's Degree in Information Technology
AI & Machine Learning major
2024 - Present
Scientific Lyceum of Information Technology, Dnipro
Official reward in my country, ALL subjects mark are 5
Diploma for innovative manifestation, leadership qualities and high level of creative performances at events
2018 - 2024

Other Experience & Activities

Self-developing and staying updated with innovations. Taking part in many cognitive activities.

Speaker, presenter, organizer. Recent examples: speaker at "Expert and Support Forum for Rebuilding Ukraine" Helsinki, "New Dimensions in Construction 2024"; workshop (presenter) Pedaforum 2025.

Conference participation: AI in AEC 2025, Slush Events, Python meetups (monthly), AWS GameDay, SRE Finland, Unity 6 stream, SoftServe.

Languages

English Fluent
Ukrainian Native
Russian Native level
Finnish Beginner

Academy of Sciences of Ukraine (МАН)

Physics Lab in the Smartphone: Learning Science in Wartime (PhyLabPhone)
First author. All-Ukrainian Competition, 2024 – Educational and Gaming Programs Category. Achieved 1st place in the national ranking and 2nd place in the final evaluation. Indie-developer.
Study of Ferromagnetic Fluid and Its Comparison with Newtonian Fluid
First author. Physics Section – Experimental Physics Category. Awarded 3rd place in the regional stage of the national competition.
Comprehensive study of the wave motion of matter
First author. Physics Section – Experimental Physics Category. Accepted in the first round of the Academy of Sciences of Ukraine (Manuscript review).

Awards & Certificates

2025
3rd Place over almost 50 teams with the game built in response to Supercell's "Context-Aware Characters". Fully developed Python backend server logics with LLM integration. Wrote C# scripts to integrate the backend inside the game, Unity3D animations setup script, UI.
Raksa AI Innovation Challenge (AI sales lead generation, internal tool in production)
2024
1st Place - Designed client- and server-side platforms with custom algorithms for autonomous client discovery and recommendations for Sales Leads generating for Metropolia's SmartLab.
All-Ukrainian Competition of Academy of Science
2024
Achieved top 1st place in all Ukraine (with 2nd place in gradation), and top-ranked in Dnipro-region and city R&D competitions with the PhyLabPhone app. Educational and Gaming Programs category.
2022 and 2024
Received Honorable Mention (Top 25%) in both Individual and Team Contests.

Technical Skills

Programming

Python Backend, automation, data processing, trading, SQL
C# (Unity)
JavaScript (& Vue.js)
Java, Pine Script, Solidity, Tact

Areas of Expertise

    AI/ML, LLMs integration: RaG/CaG, fine-tuning, CoT, Self-correction Multi-agent system design, AI vision etc. Web Development, Game Development, Automation & Bots Data Processing, Analysis and Visualization Cybersecurity & Blockchain.

Publications

Automated Toxicity Detection in Online Gaming Chats: A Comparative Analysis of NLP Methods
First author. Research paper evaluating LLMs, fine-tuned transformers, and RAG approaches for gaming chat moderation. Enterprise-grade hybrid pipeline achieving 97.5% accuracy with < 50ms processing time. Multi-stage ML pipeline: zero-tier word filtering, SBERT embeddings, fine-tuned DistilBERT (yehort/distilbert-gaming-chat-toxicity-en), and RAG enhancement. Published on PyPI, GitHub, and HuggingFace. Technologies: Python, Transformers, PyTorch, HuggingFace, PyPI packaging, CI/CD.
Evaluating OpenAI GPT Models for Translation of Low-Resource Uralic Languages: A Comparison of Reasoning and Non-Reasoning Architectures
First author. Comparative study of reasoning vs non-reasoning GPT models for translating Finnish to low-resource Uralic languages. Accepted at IWCLUL 2025 Workshop.
A Comparative Analysis of Ethical and Safety Gaps in LLMs using Relative Danger Coefficient
First author. Research; Python, LLMs, AI, Ethics. Accepted in Natural Language Processing for Digital Humanities (NLP4DH) conference in the "Long papers" category.
ORACLE: Time-Dependent Recursive Summary Graphs for Foresight on News Data Using LLMs
System that transforms daily news into weekly summary graphs using LLMs for university decision-making and foresight. Accepted at IWCLUL 2025 Workshop.