wip2025CASE STUDY 06 / 07

SkySift

India-focused aviation intelligence — live flight tracking, price discovery, AI route insights.

FastAPIPythonMapbox GLReactTypeScriptClaude APIRedis

60s

Refresh interval

6

Containers

7 days

Insight window

4

Background jobs

Screenshots coming soon

01 / The problem

Flight trackers show you dots on a map. They don't tell you whether a route is normally this busy, which airline dominates it, or when the quiet hours are.

02 / The approach

Pairs a 60-second OpenSky polling loop with proactive background enrichment — route and metadata workers populate data for every visible flight automatically, not just ones a user clicks — then a daily Claude job turns 7 days of accumulated history into a narrative per route.

03 / The catch

Insights need history to mean anything — a route's AI summary is only trustworthy after a few days of accumulated flight data, so the product is intentionally quiet on day one.

Real-time aviation intelligence platform built for the Indian market. Pulls live flight data from OpenSky Network, processes it with FastAPI + PostgreSQL + Redis, and visualises it on Mapbox GL maps. AI-powered route insights via Claude API. State managed with Zustand; deployed via Docker Compose on Coolify.

Highlights

Proactive enrichment — background workers, not click-triggered lookups

AI-generated route narratives (frequency, top airlines, peak hour) via Claude

India-focused bounding box, architected to extend globally

6-container Docker Compose stack: api, poller, cron, frontend, postgres, redis

Like what you see?

Let's build something together.