
SupplyGuard
AI-powered supply chain risk platform. Monitors disruptions, predicts bottlenecks, and flags vulnerabilities before they hit your operations.
The problem
Supply chains break in small, quiet ways long before they break loudly. By the time a shipment is late or a supplier is down, the damage is already working its way through operations. Teams need the signal early — while there's still time to reroute.
What I built
SupplyGuard watches the inputs to a supply chain and surfaces risk before it becomes an outage. An agent pipeline ingests signals — news, logistics data, supplier health indicators — and turns them into ranked, actionable alerts instead of a firehose.
Stack & decisions
- Agents: a multi-step pipeline — ingest → classify → rank → explain — so each stage can be tuned independently.
- Backend: FastAPI orchestrates the agents and serves the alerts API.
- Frontend: Next.js dashboard with a clean risk feed and drilldown views.
- Why agents, not just rules: rules miss nuance. Agents reason about why a signal matters for a specific supply chain, not just whether it tripped a threshold.
Outcome
Won the hackathon it was built for — the judges highlighted the reasoning layer as what separated it from a standard alerts dashboard. Turned a firehose of supply chain noise into a short, ranked list of "things a human should actually look at today."