- Built a data-driven Change Detection Engine and RAG-powered context pipeline to standardize and monitor six archetypes of location data.
- Developed data ingress workflows using Scrapy and Playwright, and applied LLM-based normalization with Ollama, Gemini API, and local models.
- Integrated Vector Databases (FAISS / Chroma) for retrieval-augmented generation and contextual data supply to LLMs.
- Implemented hybrid change detection logic combining rule-based methods with LLM/agent reasoning for semantic drift tracking.
- Evaluated MCP-based orchestration, Airflow automation, and agnostic-LLM benchmarking to optimize scalability and reliability.
Python
LLMs
RAG
Scrapy
Playwright
FAISS
Chroma
Airflow
Gemini API
Ollama