Projects

Developed an English-to-Portuguese NMT model using LSTM with attention in TensorFlow. Evaluated performance with ROUGE-1 metrics.

Developed a Transformer-based Question Answering system using the T5 model fine-tuned on SQuAD 2.0, a dataset for machine reading comprehension.

Developed a Transformer-based text summarization model utilizing attention mechanisms to generate summaries. Implemented inference techniques such as greedy decoding for effective word prediction in summarization tasks.

Built a Siamese network in TensorFlow to detect duplicate Quora questions. Used LSTM layers, triplet loss, and cosine similarity for training and evaluation.

Developed a neural network to recognize handwritten digits (0-9) using TensorFlow, implementing ReLU and Softmax activation functions for multiclass classification.

Developed a job search platform with my teammates enabling user authentication, and profile management, including seamless job posting and application features for companies and applicants.

Designed and implemented an enhanced single-cycle MIPS processor with a versatile instruction set. Developed the CPU using Verilog HDL, including modules for instruction decoding, ALU operations, and memory access.

Developed a Unix shell in C, featuring command execution, foreground/background processes. Implemented I/O redirection and a bookmarking system for saving, listing, and executing commands.

Collaborated on a three-iteration project for a course registration system, starting with java and concluded with a python transition. Updated artificats, including RAD and DCD, across iterations. Applied object-oriented design principles and led the implementation of critical classes in each iteration.

Collaborated with team members to implement a comprehensive database system for a herb seller business market named Ortaköy Aktarı, designed Enhanced Entity Relation (EER) tables. • Utilized MSSQL Server to manage customers, orders, and product data, ensuring efficient and secure data handling.