Co-authored a paper introducing MoneroVis, an interactive visual analytics system for exploring and tracing transactions in privacy-preserving blockchain networks.

Designed and built the platform for the Monero blockchain, utilizing state-of-the-art visualization techniques to expose underlying cryptographic structures, and implemented a custom high-performance indexing engine to map ring signatures in reverse, enabling real-time 'forward-looking' traversal of the transaction graph.

Leveraged this architecture to implement advanced heuristics, such as automatic tracing of poisoned outputs, to analyze monetary flows within the network.

Demonstrated system efficacy through a forensic case study tracing the WannaCry 2.0 ransomware funds and a domain expert study evaluating analytical confidence in privacy-preserving networks.

Developed a full-stack LLM chat platform with Spring Boot and Angular, using Spring AI to manage local LLM inference via llama.cpp.

Implemented branching conversation flows, persistent chat history, multimodal ingestion, and rich Markdown, LaTeX, and code rendering, alongside RAG pipelines powered by pgVector for semantic search.

Built a responsive UI with real-time generation metrics and seamless mid-conversation model switching.

Built a self-hosted, non-custodial point-of-sale (POS) system for accepting Monero payments, integrating directly with monero-wallet-rpc.

Designed a full payment lifecycle including invoice creation, subaddress generation, real-time status tracking, and webhook-based backend integration.

Implemented fiat-to-XMR conversion, underpayment handling, invoice expiration, and background workers for blockchain polling and price updates, paired with a lightweight embeddable checkout widget and browser-based POS interface for both in-person and online use.