Hi, I’m Aydin Aghaliyev — a Data Scientist & AI Engineer based in Rotterdam, The Netherlands.
At Philips, I build agentic AI systems for root-cause analysis and contract reasoning, combining knowledge graphs, MCP, and cloud infrastructure. My personal projects explore machine learning, quantitative trading, and GPU programming.
Projects
Variational Autoencoder
Dual-modality VAE in TensorFlow/Keras trained on MNIST images and character-level text, with KL annealing, free-bits regularisation, and an interactive latent-space exploration UI built with Streamlit.
Sentiment-Based Trading Strategy
FinBERT-based trading algorithm that analyzes sentiment from financial news headlines to generate trading signals.
GPU Mandelbrot Renderer
Mandelbrot set renderer using Apple Metal compute shaders in C++, with per-pixel evaluation parallelised across GPU thread groups via metal-cpp.