Login- Consulting
- Training
- Partners
- About Us
x
IT Services and IT Consulting
AWS Lambda, AWS IoT Core, Amazon API Gateway, Amazon SQS, Amazon S3, Amazon DynamoDB, Amazon Bedrock
Built a GenAI smart cart pipeline achieving 98% product identification and 90-95% OCR accuracy on AWS
The client is a software company established in 2017 that solves supply-chain and retail operation challenges using AI and mobile edge computing solutions. Recognized under DPIIT’s Startup India program, their platform supports inventory, payments, operations, and reporting, serving over 5 lakh businesses to streamline operations and reduce manual effort.
Product Identification Accuracy
OCR Accuracy
Rule-Based Billing Validation
The client needed highly accurate product recognition for smart carts despite inconsistent camera angles, lighting conditions, and packaging variations. The system required multi-image validation, real-time OCR and weight verification with ±10g tolerance, duplicate prevention to avoid race conditions, low-latency billing updates during shopping, and scalable data ingestion from field devices without disrupting the shopping experience.
• Field devices upload images to S3 and weight/session data to IoT Core (MQTT)
• Amazon S3 triggers Amazon SQS, which invokes Lambda for controlled parallel processing
• Lambda preprocesses images and uses Bedrock Claude 3.5 Sonnet for OCR and label extraction
• OCR results sent to Amazon EC2 FastAPI, which uses Claude 3 Haiku + Bedrock Knowledge Base + Titan Embeddings for product identification and SQL generation
• Product validated using cart weight from Amazon DynamoDB; bill updated in Amazon DynamoDB
• Mismatches trigger IoT drop alerts back to the device
Delivered a GenAI smart cart pipeline achieving 98% product identification, 90-95% OCR accuracy, and enhanced billing precision through rule-based validation.
SHARE
Our support doesn't end here. We have monthly newsletters, study guides, practice questions, and more to assist you in upgrading your cloud career. Subscribe to get them all!
