NetworkMind

Context

NetworkMind combines professional contact data, public web research and LLM-assisted responses to help users form a clearer view of relevant people in their network.

Problem

Professional search becomes noisy quickly when contact data, public research and generated answers are handled separately.

What I built

A prototype flow that routes questions to relevant AI and search tools, combines retrieved information and supports contact prioritisation through ranking and validation paths.

Technical approach

• Structured the prototype around tool selection and fallback behavior so failures could be handled deliberately.
• Worked on prompt flows and retrieval logic to keep generated responses tied to available research signals.
• Tested common routing and response failure scenarios while implementing full-stack pieces across the application.

What I learned

The project sharpened the importance of clear routing rules, observable fallbacks and product-shaped validation when AI is part of an application workflow.