Telecommunications and Signal Processing engineer with a strong interest in artificial intelligence, machine learning, and building reliable AI-powered products.

I enjoy working across the full development cycle, from data processing and model evaluation to backend APIs and user-facing applications.

I'm currently developing BankScope, an evidence-grounded RAG assistant for researching technology, cybersecurity, and operational risks disclosed in U.S. bank filings.

The project combines:

• automated SEC filing acquisition and structure-aware document processing;
• dense and lexical retrieval with Qdrant, BM25, and reciprocal rank fusion;
• citation-backed answer generation and multi-bank comparisons;
• conversational memory with strict evidence boundaries;
• retrieval and generation evaluation pipelines;
• a FastAPI backend and React/TypeScript frontend.

A major focus of the project is making AI answers traceable, measurable, and grounded in source documents rather than relying on unsupported model knowledge.

• Retrieval-Augmented Generation and information retrieval
• Large language model applications and evaluation
• Deep learning and representation learning
• Computer vision and multimodal systems
• Signal processing and wireless communications
• AI applications in banking and operational risk
• Building end-to-end machine learning products

Some of the projects and topics I have explored include:

• an end-to-end autoencoder for wireless communication systems;
• visual language identification from silent lip movements;
• learned image compression;
• computer vision classifiers;
• neural networks for signal-processing problems.

One of my favorite projects is a classifier that distinguishes Serbian from English using only lip movements—without audio. It achieved approximately 65% accuracy against a 50% baseline.

I'm interested in collaborating on:

• practical AI and machine learning applications;
• RAG and semantic-search systems;
• computer vision or signal-processing projects;
• Kaggle competitions;
• open-source tools for evaluating and improving AI systems.

Feel free to ask me about BankScope, retrieval systems, deep learning, lip reading, learned image compression, or signal processing.