Software / AI Engineer building backend services and LLM-powered systems with Python and FastAPI — REST APIs, RAG and semantic search, OCR pipelines, vector and graph databases.
Finishing my B.Eng. in Electronics & Computer Engineering at Wrocław University of Science and Technology
- Backend & API systems — Python, FastAPI, REST APIs, JWT auth, PostgreSQL, SQLAlchemy, Alembic
- AI-powered products — RAG, LangChain / LangGraph, MCP (Model Context Protocol), prompt engineering, semantic search, OCR pipelines
- Data-driven applications — PostgreSQL + pgvector, Supabase, Neo4j, Pinecone
- Computer vision & forecasting — PyTorch, OpenCV, YOLO, scikit-learn, time-series methods
-
Lucid — AI Notes Platform 📱 live in App Store beta FastAPI backend with 20+ REST endpoints (OCR upload, semantic search, AI Q&A, flashcards, quizzes), covered by 200+ unit and integration tests. Pipeline: Google Vision OCR → OpenAI embeddings → pgvector (HNSW) → LLM answers.
-
Knowledge Graph RAG / ToPWR Natural-language-to-Cypher retrieval over a Neo4j knowledge graph with LangGraph guardrails, FastAPI and an MCP server — built at Solvro for the ToPWR university app.
-
NeuroPark — 3D Vehicle Detection Single 2D traffic-camera images → 3D vehicle scenes using YOLOv8, MiDaS depth estimation and Open3D, validated against LiDAR. Presented at the KPZ25 Engineering Conference.
-
TensoCast — Time-Series Traffic Forecasting CP/Tucker tensor decomposition combined with LSTM and XGBoost on the METR-LA dataset — 35% improvement over baseline models.
🌐 canerolcay.com · 💼 LinkedIn · 📍 Wrocław, Poland