Antibody

• 

  Antibody: thousands of events in, one root cause out

Inspiration

When one microservice fails, every service around it starts alerting too. On-call engineers drown in thousands of events while the actual cause hides in the noise.

What it does

Antibody learns the dependency graph from traces, folds thousands of events into one incident and ranks the likely root cause with a confidence label. When the evidence is thin, it says so.

How we built it

Python and FastAPI, with WebSockets for live updates and a PageRank-style ranking over the learned dependency graph.

Accomplishments that we're proud of

• 1st in college at the Microsoft Codathon 2026 (problem 3), with 9,900 points in a team of five
• On an 84-scenario benchmark, the true root cause ranked first in 100% of clean runs and 84.5% in hard mode

Victor led the technical explanation in the pitch.

Built With

• fastapi
• python
• websockets