A real-time pipeline that paid for itself in six weeks.

A healthcare data company was running a near-real-time pipeline on six always-on EC2 instances behind an Application Load Balancer, processing roughly 40 events a second at peak. The bill was $8,200 a month. We rebuilt it serverless and brought it to $640.

INDUSTRY

Healthcare data

DOMAIN

COST

DELIVERED

2025

STACK

EVENTBRIDGE·LAMBDA·KINESIS DATA STREAMS·DYNAMODB·S3·STEP FUNCTIONS·X-RAY

RESULTS

What changed, by the numbers.

COMPUTE BILL

−92%

$8.2K → $640 / MONTH

p95 LATENCY

−60%

480ms → 190ms

ON-CALL PAGES

−85%

AUTOSCALES, NO BABYSIT

ROI

BREAKEVEN ON ENGAGEMENT

HOW IT WENT

The instances were sized for peak: a 7am batch from a partner that processed ~3,000 events in 90 seconds, twice the rest-of-day throughput. The other 23 hours, the fleet ran at 8% utilisation.

We replaced the pipeline with EventBridge fanning out to Lambda consumers, with Kinesis Data Streams in front for the burst-tolerant path and Step Functions orchestrating the multi-step records. DynamoDB on-demand replaced the hot Postgres tables; the cold ones moved to S3 with Athena.

The team kept their old Postgres for reporting (it cost $400/mo to leave running, and nobody had time to rewrite the dashboards). The migration ran shadow for two weeks before the cutover. No data loss, no on-call wake-ups, no surprised invoices.

RELATED · SAME DOMAIN

Other engagements in this space.

VANTAGE · 2026

A $400k monthly bill, with the right commitments for once.

37%EFFECTIVE SAVINGS

LUMEN · 2026

Model training on Spot, checkpoint-resumed.

−74%TRAINING COST

PREVIOUSTENDRIL — Incident playbooks that run themselves while you’re still putting on your laptop.NEXT TUNDRA — SOC 2 Type II evidence that gathers itself.

READY WHEN YOU ARE

Let's get your AWS bill (and architecture) in order.

The discovery call is free. You walk away with at least one concrete idea — even if we never work together.

Or email directly →