TechPrufer
Get started
Instacart
Instacart logoInstacart · Feature trail

Caper Data Flywheel and Observability

1 post · Feb 2026

What this trail covers

Instacart built Capsight, an end-to-end platform for Caper smart carts. It acts as a distributed data collection and model improvement engine. The system comprises the Capsight Collector (on-device intelligent data capture), Capsight Depot (cloud platform for data management, ingestion, curation, and AI-assisted annotation), and Capsight Learner (distributed training platform). This enables a continuous feedback loop from edge-captured data to ML training, significantly reducing model iteration time and annotation costs through VLM pre-labeling. The system is expanding to full sensor fusion for a foundation model understanding store environments.

All Instacart trails
  1. Turning Data into Velocity: Caper’s Edge and Cloud Data Flywheel with CapsightThis post details the implementation of Capsight, an end-to-end platform for Caper smart carts. It elaborates on the Capsight Collector's trigger-based capture and optimized on-device processing for multimodal data, the Capsight Depot's ingestion, indexing, search, visualization, and AI-accelerated annotation using a VLM-based pre-labeling service (projected to reduce annotation costs by over 70%), and the Capsight Learner's Ray-based distributed training platform that reduced the model training stage from one week to two days. The post highlights the impact of these components in improving model accuracy by over 5% and accelerating the iteration cycle from a month to a week. Future work includes expanding to full sensor fusion for a foundation model.