Cyber Intel Pipeline
Built a decoupled Scraping - Normalization - Analysis pipeline where Selenium simulates human browsing, heuristic parsing converts noisy HTML into structured JSONL, and downstream AI only consumes high-signal threat data.
- Uses source grounding and strict JSON schemas so every reported risk remains verifiable and actionable.
- Balances local privacy with scale through Ollama and Gemini-based summarization paths.