Projects I Built/Contributed To:1. Similar Image Search: - Developed FastAPI endpoints. - Tools: Python, FastAPI, Docker, TF Serving, ElasticSearch. 2. Decompounding: - Segmented compound words into constituent parts. - Tools: Python, spaCy, stanza. 3. Domain-Specific Synonym Generation: - Tools: Python, word-embeddings, TF Serving, FastAPI, Docker.4. Knowledge Graph Construction: - Built a system for centralized data storage and inferencing. - Tools: TypeDB, Python, FastAPI, Docker.5. Catalog Domain Classification: - Tools: Python, scikit-learn, Docker.6. Semantic Search Engine Development: - Integrated OpenSearch with FastAPI; matched queries using BM25 and sentence embeddings. - Tools: Python, OpenSearch, FastAPI, NLTK, SentenceTransformers, Docker.7. Intent Classification: - Trained models with Rasa for intent recognition. - Tools: Rasa Python SDK, FastAPI, Docker.8. Complete the Look Prompts: - Formulated LLM prompts for suggestions. Created a Streamlit app for QA comparisons. - Tools: Prompt Engineering, Python, Streamlit.9. Nvidia-Triton-Server Deployment: - Prepared deployment for image captioning. - Tools: Python, Nvidia-Triton-Server, Transformers.10. Prompt Development for E-Commerce Tasks: - Extracted main product nouns and classified products. - Tools: Prompt Engineering, Python.

- Spearheaded the development and implementation of NLP models for automated tagging within the DEEP framework, tailor-made for secondary data analysis.- Managed the complete lifecycle of model development: from data acquisition and preprocessing, to the design, training, and testing of ML models, culminating in their production readiness.- Proactively optimized models in response to user feedback, underscoring a dedication to adaptability and a relentless pursuit of enhancement.

Spearheaded Machine Learning initiatives for forum.donanimhaber.com, focusing on:- Design and implementation of a Recommender System to enhance user experience and content relevance.- Text Classification techniques to streamline content categorization and management.- Implementing and refining Ranking algorithms for improved information retrieval.

Project: Social Media Trend Detection and AnalysisDuring my master's studies, separate from my thesis, I played a significant role in a three-part project in collaboration with the Technology Transfer Office. This project encompassed:• Named Entity Recognition (NER): This entailed identifying and classifying named entities within text to derive structured information.• Sentiment Analysis: The objective was to analyze and identify emotions and sentiments expressed in social media content, thus gauging public perception.• Trend Detection: I was involved in developing offline trend detection models.In each segment of the project, my responsibilities included:- Data Collection: Sourcing and compiling relevant datasets for model training and evaluation.- Data Cleaning: Ensuring the integrity of data by eliminating noise, addressing missing values, and standardizing inputs.- Labeling: Categorizing data points to support supervised learning and enhance model precision.- Model Development: Creating and training models tailored to the specific needs of each project segment.