Building What Buffett Said? — a retrieval-based system to search and answer questions over Buffett's shareholder letters 📜

Splitting long financial documents into smaller sections and using embeddings to retrieve relevant information.

#LangChain #RAG #OpenAI #Python

Built a GNN model with a prototype-based layer to diagnose patients from electronic health record (EHR) graph data 🧠

Adapted an existing prototype-based approach to healthcare data (MIMIC-III), identifying the parts of the patient graph influencing each prediction.

Evaluated model performance and compared results with common explanation methods.

#GNN #XAI #Healthcare #MachineLearning

A collaborative document editor built for students and teams, with real-time editing, note sharing, and no tracking 🚀

Designed as a simpler, privacy-focused alternative to mainstream editors.

#OpenSource #Collaboration #WebDev

Analyzed how different asset classes — stocks, bonds, real estate, gold, and bitcoin — performed as inflation hedges in Turkey between 2018–2024 📊

Found that most conventional hedges underperformed during periods of high economic instability.

#DataScience #Economics #Turkey

Three projects from CS464 – Introduction to Machine Learning 🎓

📌 Naive Bayes text classification — MLE vs MAP estimators with Dirichlet priors on imbalanced data
📌 PCA image compression + logistic regression — stochastic, mini-batch and full-batch benchmarks
📌 Second-hand car price prediction — linear models, random forest, SVR & a tuned DNN

#MachineLearning #NaiveBayes #PCA

Four projects from GE461 – Introduction to Data Science 🧪

📌 Dodgers attendance — linear regression on promotions, weather & opponents; bobblehead nights matter!
📌 MNIST dimensionality reduction — PCA & Isomap + QDA, visualized with t-SNE
📌 ANN hyperparameter effects — how learning rate, hidden units & normalization shape loss curves
📌 Concept drift ensemble — ADWIN-monitored adaptive model switching on streaming data

#DataScience #R #StreamingData