01 / About
I'm Berat Mert Kayacan, an Artificial Intelligence Engineering student at Bahçeşehir University, pursuing a double major in Management Engineering. I work at the point where machine learning and data science meet real operations: exploring and modeling data, forecasting demand, optimizing inventory, and auditing models for fairness and explainability.
University
Bahçeşehir University
Major
AI Engineering
Double Major
Management Engineering
Based in
Istanbul, Turkey
02 / Experience
Responsible AI Intern
Data Analytics & Operations Intern
Sponsorship Committee Associate
Managing brand outreach and sponsor acquisition, building and maintaining corporate partnerships.
Organization Committee Associate
Coordinating event operations from preparation through event day.
03 / Projects
Full-Stack · AI · Inventory
An AI-powered inventory management platform. Foresto tracks every product and stock movement in real time, uses machine learning to predict how much of each item will sell in the coming period, and recommends exactly when and how much to reorder.
Hackathon · FinTech · Mobile
A personal finance assistant. Add your bank statement as a PDF and Finnalyze reads it for you: every transaction is categorized automatically, installments and debts are tracked in one place, and the assistant comes back with personalized advice on spending, saving and paying off debt. Built with Flutter and Google Gemini at BTK Akademi Hackathon 2026.
Home
Documents
Analysis
Debt Plan
AI Recommendations
Statistics · R · Research
Statistical analysis of 22,865 Inside Airbnb listings in Istanbul (15 variables): frequency distributions, central tendency, skewness, district and room-type breakdowns, and Europe vs. Anatolia shore comparisons using R.
Research question: Which factors drive Istanbul Airbnb nightly prices, and do listings differ meaningfully by district, room type, and geographic shore?
Observations
22,865
Variables
15
Min Price
₺80
Max Price
₺7,787
Product · Marketing · D2C
Releaf is a plush Pomodoro desk companion that helps you stay focused, and an end-to-end marketing project built from scratch. I owned the full go-to-market cycle: market research and segmentation, student-focused targeting and positioning, brand identity, pricing strategy, social media campaigns and ad creatives, and a direct-to-consumer storefront at releaftr.com.
Statistics · Simpson's Paradox
Restaurant chain branch manager comparison using SQL, pandas and statistical tests, exposing how aggregate satisfaction rates reverse when segmented by location type (mall vs. industrial).
Management saw Branch B at 77.4% vs Branch A at 53.9% and recommended a manager change, but when location type is held constant, Branch A wins in every segment.
Branch A (aggregate)
53.9%
Branch B (aggregate)
77.4%
Branch A in malls
92.5%
Branch A industrial
44.3%
A high-performance C++ engine built to crunch huge CSV files. Fed with 5 million NYC taxi trip records, it parses, indexes and answers analytical questions like the busiest zones and peak hours in under two seconds, using a custom composite hash index and careful memory management instead of heavyweight frameworks.
+ More on GitHub
Analytics · Statistics
A collection of small, standalone data science studies:
Time Series · ML
End-to-end time series forecasting on Kaggle's real-world Favorita dataset: sales records from thousands of product families across stores in Ecuador, with dates, promotions and store metadata. Extensive EDA, feature engineering and ML models.
Machine Learning · R
Comparative classification study in R: KNN, Decision Tree and SVM models trained on a simulated customer purchase dataset, with feature scaling and model evaluation on a shared test set.
04 / Skills
Models that matter
Python, scikit-learn, demand forecasting & EOQ modeling, regression & classification (KNN, SVM, decision trees), K-Means & PCA, LLM integration with the Gemini API
AI you can trust
Explainability (XAI), fairness and robustness for ML systems, Infosys Responsible AI Toolkit, LLM explainability experiments
From raw to insight
Pandas, NumPy, EDA & feature engineering, statistical analysis & hypothesis testing, SQL, C++ for high-performance data processing, Matplotlib & Seaborn
Numbers meet strategy
Project management, digital marketing, go-to-market strategy, operations research, strategic planning
The daily stack
FastAPI, Git & GitHub, Excel, AI-assisted development (Cursor, Claude)
+ Approach
I combine AI engineering with management principles, from demand forecasting and Responsible AI to go-to-market strategy and operational optimization. Every project starts with data, ends with impact.
Istanbul, Turkey · beratmert.kayacan@gmail.com
