Context
Argus began as an attempt to make live video review feel more humane for teams that need signal without sitting through every frame themselves.
The central question was whether ingest, computer-vision analysis, and operator review could read as one coherent workflow instead of three disconnected tools.
Build
I framed the demo around one continuous loop: bring in a live feed, analyze it automatically, and surface notable events in a review interface that still felt operator-led.
The prototype stayed grounded in real constraints, with streaming, typed application code, and persistent storage carrying the idea beyond a superficial hackathon mock-up.
Takeaways
The result suggested that computer-vision assistance could reduce constant manual watching while still preserving clear review points for a human operator.
It also became a stronger portfolio story because the architecture, product choices, and tradeoffs could be explained as one end-to-end system.