← Home
The Company

Making existing cameras intelligent.

Most buildings already have cameras. Almost none of them have anything that understands what's in frame. QuantumSight is the layer that does — built on top of the CCTV that's already there.

The problem

Most buildings already have cameras. What they don't have is anything that understands what's in frame — so footage only matters after something has already gone wrong.

QuantumSight sits on top of existing CCTV — no rip-and-replace. An edge box runs vision models on the local feed and turns raw footage into events a security team can actually act on.

CAM_11 · DOCKLIVE
PERSON · 0.94
BAG · 0.71
How it works

Four steps between a camera and something worth a person’s attention.

  1. 01
    Existing CCTV

    Whatever's already on the wall. RTSP in, no new hardware on the camera side.

  2. 02
    QS AI Box

    An edge device running vision models on the local feed. Nothing leaves the building unless it matters.

  3. 03
    Cloud inference

    Heavier models and cross-camera context for the events that need a second look.

  4. 04
    Alerts dashboard

    What a security team actually watches — events, not 40 live tiles.

Where it is now
CONCEPT
PROTOTYPE
PILOT
SCALE

Pipeline: existing CCTV → QS AI Box (edge) → cloud inference → alerts dashboard.

TODO — a paragraph on what you're building right now and what you've learned shipping it.

BUILT WITH — COMPUTER VISION · EDGE INFERENCE · PYTHON · REACT

Build log

Short notes from building in public. Newest first.

TODO — drop dated entries here (content/quantumsight-log.ts). One or two lines each: what shipped, what broke, what you learned.

Have cameras that could be doing more?

We’re running early pilots. If you operate a site with existing CCTV and a security team that’s drowning in tiles, let’s talk.

saksheth.rao@gmail.comSee other builds →