Junior Division
High school students
For students building strong foundations in Python, data, and ML basics.
Balkan AI & Machine Learning Olympiad
An individual regional competition for students who can turn messy data into working AI solutions, explain their choices, and think clearly about responsible use.
- Finalists
- 147
- Countries
- 12
- Individual rounds
- 5
- Divisions
- 3
18-20 June 2026
Sofia Tech Park, Sofia
About
A real regional stage for individual AI talent
The olympiad is built around the work students actually need to do in AI: understand a problem, inspect data, choose a model, test it honestly, and explain the limits.
Students compete alone. National qualification rounds can feed into the final, and direct applications are accepted from countries still building a local AI competition pathway.
The final is not a programming sprint or a leaderboard race. It combines olympiad-style reasoning, practical ML notebooks, applied regional tasks, and responsible AI review.
Senior Division
University students and advanced high school students
For competitors ready to handle deeper modeling, validation, and analysis.
Open Division
Self-taught participants and independent learners
For young developers who prepared outside a formal school track.
Results
2026 winners and special awards
Results are published as a news article so the homepage stays focused and updates live on their own pages.
1st
Stoyan Ivanov
Bulgaria - Gold Medal and Overall Champion
Top combined score across the ML challenge, code review, and explanation round.
Score
91.4
2nd
Elena Popescu
Romania - Silver Medal
Strong data analysis notebook and a stable model validation strategy.
Score
88.9
3rd
Nikola Petrovic
Serbia - Bronze Medal
Excellent Python implementation and one of the clearest error analyses.
Score
86.7
Best Machine Learning Solution
Mira Kostic, Croatia
Best Responsible AI Analysis
Maria Papadopoulou, Greece
Best Beginner Performance
Arda Yilmaz, Turkey
Best Real-World Impact Project
Luka Markovic, Montenegro
Format
Five rounds, one competitor per submission
Each round checks a different part of practical AI work, from theory to production-minded explanation.
R01
Theory Round
ML concepts, probability, statistics, neural networks, evaluation metrics, and ethics.
R02
Coding Round
Python problems using data structures, algorithms, NumPy, pandas, and simple ML implementations.
R03
Machine Learning Challenge
A shared dataset where each participant trained, validated, and submitted a model independently.
R04
Applied AI Task
A real regional problem such as air quality, public transport, satellite imagery, or misinformation.
R05
Explanation Round
A short written and oral defense covering assumptions, limitations, ethics, and real-world use.
Schedule
Three days from check-in to jury moderation
The final is structured like a competition event, with fixed blocks, published checkpoints, and a clear awards window.
Day 1
18 June
09:00Registration and technical check
10:00Opening briefing and rules
11:00Theory round
14:30Python coding round
Day 2
19 June
09:30Machine learning challenge released
12:30Data analysis checkpoint
15:00Applied AI task
18:00Submission freeze
Day 3
20 June
09:30Explanation round
12:00Responsible AI review
14:00Jury moderation
16:00Awards ceremony
Scoring
Accuracy matters, but it is not the whole result
A strong AI olympiad rewards complete thinking. Competitors have to show why their model is valid, readable, and responsible.
Model performance
35%Accuracy, robustness, validation, and metric choice.
Code quality
20%Readable notebooks, reproducibility, clear structure, and clean Python.
Data analysis
15%Cleaning, features, visualization, leakage checks, and assumptions.
Explanation and reasoning
15%Why the method fits the task and what the model cannot safely claim.
Ethics and limitations
10%Bias, privacy, misuse risk, transparency, and deployment boundaries.
Creativity
5%Original feature ideas, simple improvements, or useful product framing.
Competition tasks
Regional data, practical constraints, clear explanations
Tasks are designed around problems students can understand and communities can recognize.
Predict PM2.5 and NO2 levels using weather and sensor data from Sofia, Belgrade, Bucharest, and Athens.
Detect misleading regional news articles and explain why a classifier may be biased.
Classify satellite images for wildfire and flood risk across mountain and coastal areas.
Forecast municipal energy consumption during heat waves and major public events.
Analyze public transport delays and identify which model errors matter to passengers.
Design a small public-service AI assistant and document safe-use boundaries.
Skills
What competitors are expected to know
The field spans programming, data science, ML theory, applied modeling, and responsible AI communication.
PythonNumPypandasscikit-learnStatisticsFeature engineeringNLPComputer visionModel evaluationAI ethics
News
Latest updates
Results20 June 2026
Stoyan Ivanov wins the 2026 Balkan AI & Machine Learning Olympiad
The final scoreboard is now published after jury review of models, code, data analysis, explanation, ethics, and creativity.
Competition19 June 2026
Finalists work with regional air-quality and transport datasets
The main challenge used messy, realistic datasets with missing sensors, uneven city coverage, and time-based validation requirements.
Responsible AI18 June 2026
Ethics review becomes a deciding factor in the final score
The final review rewarded competitors who documented bias, privacy, misuse risks, explainability, and the limits of their models.
Partners
Organized with schools, universities, mentors, and data partners
A competition site should show the ecosystem behind the event, not only describe the challenge.
Balkan AI Education Network
Sofia Technical University AI Lab
Regional Data Commons
School of Digital Sciences
Open ML Mentors
Youth Tech Forum