I build end-to-end software and hold an M.Sc. from Tel Aviv University, where I was mentored by Dr. Mahmood Sharif. My research experience is in machine learning, AI, and cybersecurity.
I'm a software engineer with experience in backend development, REST API servers, cloud infrastructure, and applied machine learning. I enjoy building systems end to end, with an emphasis on keeping them practical and secure.
My graduate work focused on machine learning and artificial intelligence, specifically the use of Large Language Models to forecast cyber incidents across organizations.
I developed a retrieval-based framework using FAISS indexes and RAG-style pipelines to retrieve relevant news and geopolitical reporting, then combined LLM-based ingestion with structured network indicators to forecast security incidents at the organizational level.
This work also led to a paper currently under review, Forecasting Security Incidents Using Geopolitical Data, co-authored with Abed Garra.
In my free time, I research vulnerabilities in web applications, APIs, and exposed data paths, and report findings through coordinated disclosure.
Focused on machine learning and AI, including thesis work on retrieval-based LLM pipelines and structured network features for forecasting cybersecurity incidents.
Strong foundations in algorithms, data structures, operating systems, and software engineering.
My M.Sc. thesis focused on predicting cyber incidents using deep learning over both unstructured text and structured network features.
The system retrieved relevant geopolitical reporting through FAISS and RAG-based pipelines, processed large text collections with LLM-based components, and combined those signals with structured indicators to forecast incidents at the organizational level.
An agentic LLM-powered security scanner that analyzes HAR files extracted from HttpToolkit and generates attack hypotheses for each request.
It uses LangGraph to orchestrate request mutation and execution, then applies an LLM to review responses for false-positive filtering and vulnerability classification.