Back to Portfolio
Network Intelligence

Network Intelligence System

Physics-inspired professional network analysis. Three implementations sharing one scoring methodology: mass, velocity, priority. Built to answer which relationships matter most right now.

Physics-inspired scoring 3 Implementations Privacy-first
Professional networks decay silently. The people who can open doors drift away because nobody tracked the relationship momentum. You notice it when you need an introduction and realize you haven't spoken in 18 months. KNIS applies physics to professional relationships: mass (seniority, influence), velocity (recency, interaction frequency), and a blended priority score. The result is a ranked list of who to contact today, not who you connected with in 2019.

Scoring Methodology

Priority = 0.6 × Mass + 0.4 × Velocity
Mass Seniority and influence score. Director-level = high mass. Junior IC = low mass. Role title + company size + industry weight.
Velocity Recency and interaction momentum. Connection 3 weeks ago = high velocity. Connection 3 years ago with no interaction = near zero.
Priority Blended score. High mass + high velocity = people to contact today. High mass + low velocity = relationships worth reactivating.

Three Implementations

v1 — Standalone
Python Analyzer

Single-file Python tool. Takes LinkedIn CSV export, outputs ranked connection list with mass/velocity/priority scores. matplotlib + pyvis graph visualization.

v2 — Full-stack
SvelteKit + FastAPI

Full platform with persistent SQLite storage, REST API, and SvelteKit frontend. Upload CSV, explore graph interactively, filter by industry and seniority.

v3 — Alpha
Geocoded Maps

Alpha with MapLibre geographic visualization. Geocoded company locations across three tiers (exact, city, country). See where your network is concentrated globally.

Technology Stack

Python SvelteKit FastAPI SQLite MapLibre GL pyvis matplotlib Turborepo + pnpm