Satellite Orbit Data Pipeline Development

I’m putting together an end-to-end pipeline that ingests public TLE feeds, propagates orbits with SGP4, flags close approaches, evaluates collision probability, and serves the results through a FastAPI backend and a lightweight web dashboard. The entire focus is on timely, reproducible data analysis, so clean code, scientific rigour, and quick turnaround are essential—I need the first usable build as soon as possible.

Scope in a nutshell
• Data Collection – pull fresh TLEs from Celestrak and Space-Track, handle authentication where required, and archive raw files.
• Orbit Engine – SGP4-based propagator that can give satellite positions at arbitrary epochs with millisecond accuracy.
• Conjunction & Collision – identify conjunctions, compute probability of collision, raise alerts once a configurable threshold is breached.
• Manoeuvre & Blind-Spot Detection – ML model (your choice of framework) to spot unexpected orbital changes or loss of tracking.
• Re-entry Tracker – predict decay events and potential impact zones.
• Backend API – FastAPI service with JWT login, role-based access, and endpoints for all computations.
• Operator Dashboard – minimal React or Vue front-end that visualises alerts, pass tables, and risk scores.
• Deployment – scripts and docs for a 7-instance production setup (workers, API, db, web) on Docker or Kubernetes.

Acceptance criteria
1. Each module must include unit tests and a short README describing assumptions and usage.
2. End-to-end test proving TLE ingestion → alert visible on the dashboard.
3. Containerised deployment spins up locally with one command.

When you reply, highlight relevant experience on orbit determination, FastAPI, and ML anomaly detection—concise examples are enough. I can iterate fast, so expect prompt feedback and code reviews as we go.