Hi,
I'am Ekin
AI Engineer

Academic Projects

• UAV Trajectory Generation and Classification (Master Thesis)

  Part of European Union project called BUBBLES

  Explored various generation methods via Machine Learning models and designed a novel classification technique for UAV trajectory types.

• RAG Chatbot API for Product Information

  Implemented a Retrieval-Augmented Generation (RAG) chatbot API that allows users to interact with product data. The chatbot generates responses based on pre-calculated document embeddings and can answer queries related to product specifications, user manuals, warranties, and installation guidelines.

  This chatbot API integrates with a FastAPI backend, processing up to 5-6 PDFs per product. It effectively handles embeddings to generate accurate, real-time responses to customer queries using previously indexed product documents.

  The solution improves customer support by providing precise and context-aware answers to product-related questions.

• LLaMA Chatbot fine-tuning

  Fine-tuned a LLaMA chatbot using my personal Facebook and WhatsApp data, which consists of over 10 years of conversations.

  View Code

• Memory Game on the Pepper Social Robot

  Designed and implemented a simple memory game on the "Pepper" social robot

  View Code

Named Entity Recognition and Semantic Role Labeling with Transformers

Developed models for Named Entity Recognition and Semantic Role Labeling using modern transformer architectures

View Code

Image Segmentation with Mask R-CNN and Video Classifier

Implemented an image segmentation pipeline using Mask R-CNN on video data and integrated a domain-specific video classifier to enhance object recognition accuracy.

Web-Based Video Game with WebGL

Designed and developed an interactive web-based video game using WebGL technology, focusing on immersive graphics and seamless user experience.

View Code

Explainable Image Classification on MNIST with Integrated Gradients

Implemented a neural network for image classification on the MNIST dataset, integrating gradient-based methods for explainability.

Atari Gameplay with Deep Reinforcement Learning

Applied deep reinforcement learning techniques to enable an AI system to play Atari games, demonstrating practical applications in AI.

Venice Boat Classification

Developed a machine learning model for the classification of boats in Venice using image data, employing state-of-the-art algorithms for high accuracy.

• Guide My Boat

  A navigation system designed for boats in the Mediterranean, facilitating safer and more efficient maritime travel.

Super-Resolution Application with Residual Dense Network

Implemented a super-resolution application using a Residual Dense Network in the domain of deep learning