Indian banks sit on decades of financial records that are effectively unsearchable. Knotorious needed a system that could ingest, index and serve half a terabyte of it without the six figure infrastructure bill that usually comes attached.
We designed and built the backend: a Golang ingestion pipeline writing to S3 with lifecycle tiering, an OpenSearch index for full text and faceted queries, and PostgreSQL for relational integrity. Everything ships as Docker images through GitHub Actions onto EC2 behind Nginx.
What we built
01Designed an S3, OpenSearch and PostgreSQL storage tier that keeps queries fast while cold data moves to cheaper storage classes.
02Held total AWS spend under 1,500 dollars a month using Terraform managed modular scaling and right sized instances.
03Shipped zero downtime releases through containerised deploys and a pipeline with automated health checks.
04Hardened access with least privilege IAM roles, signed URLs and audit logging to pass banking review.
