I am an AI & Machine Learning Engineer specializing in Agentic RAG Architectures, LLM Fine-Tuning, and Enterprise Document Intelligence Systems.

I hold a Master’s Degree (M.Sc.) in Information and Data Systems from Hacettepe University. My focus is on transforming cutting-edge AI research into production-grade, high-throughput microservices — handling everything from multi-modal document ingestion (VLM/OCR) to dual-channel hybrid vector search and ReAct planning agent loops.

• 💻 Role: Senior AI & Machine Learning Engineer / Systems Architect
• 🎓 Education: M.Sc. in Information & Data Systems — Hacettepe University
• 🛠️ Core Expertise: Agentic RAG, Multi-Vector Search (Qdrant), LLMs & Transformers, High-Throughput Pipelines (FastAPI, Celery, AsyncIO)
• 🌍 Goal: Building state-of-the-art, scalable AI products and intelligent document platforms

┌─────────────────────────────────────────────────────────────────────────────┐
│                          AGENTIC RAG ARCHITECTURE                           │
├──────────────────────────────────────┬──────────────────────────────────────┤
│  • Dual-Channel Hybrid Retrieval     │  • Multi-Modal Ingestion (VLM / OCR) │
│  • Deduplication & Ranking Engine    │  • Document Summarization Vector DB  │
│  • ReAct Agent Loops & Deep-Dive     │  • RBAC Document Access Control      │
└──────────────────────────────────────┴──────────────────────────────────────┘

• 🤖 Agentic RAG & LLM Orchestration: Designing multi-stage retrieval pipelines combining Unrestricted Vector Search with Summary-Guided Filtering and intelligent deduplication.
• 📄 Multi-Modal Document Intelligence: Building end-to-end OCR/VLM ingestion workflows, converting unstructured PDFs/DOCX/Excel/CSVs into structured Markdown, tables, and multi-vector embeddings.
• ⚡ Scalable AI Infrastructure: Deploying distributed microservice ecosystems powered by FastAPI, Celery, Qdrant, PostgreSQL, Redis, and Docker.

• Description: An end-to-end enterprise RAG platform designed for complex multi-page document querying, RBAC access control, table discovery, and multi-modal document analysis.
• Key Features: Dual-channel hybrid retrieval engine (unrestricted + summary-guided with deduplication), ReAct agentic reasoning loops, VLM/OCR ingestion pipeline, and high-performance asynchronous Celery workers.
• Tech Stack: Python, FastAPI, Qdrant Vector DB, PostgreSQL, Redis, NestJS, React, Docker.

• Description: End-to-end NLP system for abstractive text summarization and rating-conditioned review generation.
• Key Features: Transformer fine-tuning using T5 for abstractive summaries, DistilGPT-2 for rating-conditioned generation, and embedding-based rating prediction with interactive Streamlit interface.
• Tech Stack: PyTorch, Hugging Face Transformers, T5, DistilGPT-2, Streamlit.

• Description: Comprehensive repository containing production-level machine learning experiments, deep learning model training scripts, feature engineering workflows, and NLP evaluations.
• Tech Stack: Python, PyTorch, Scikit-Learn, Pandas, NumPy.

• Description: Enterprise relational database system featuring relational modeling, indexing strategies, and optimized backend transactional query execution.
• Tech Stack: PostgreSQL, SQL, Database Design & Optimization.

• 💼 LinkedIn: omercansagbas
• 📧 Open to discussions around Machine Learning, Large Language Models, RAG Architectures, and Applied AI Engineering.