Case Study • Real-time Video + Telemetry

RaceKeeper Connect

A production-grade AWS pipeline for dual-camera live streaming and real-time telemetry, including automated provisioning and a HUD-style web player.

Discuss a similar build Cloud Solutions FAQs

AWS • Kinesis Video Streams Telemetry + HUD Overlay Automated Provisioning

Objective

Deliver a reliable live experience that combines two camera angles (front + rear) with real-time telemetry for motorsport sessions, and make it operationally scalable with automated provisioning.

Outcome

Dual-camera live streaming with canonical HLS endpoints per device
Telemetry ingestion pipeline producing a “latest.json” feed
HUD overlay player with data freshness/timeout logic
Provisioning automation to onboard devices consistently

What made this hard

Live video stability, encoder settings, stream availability, and telemetry freshness all have to work together. A “works on my machine” setup breaks quickly at scale.

Per-device stream naming and multi-region constraints
Dual-stream orchestration (front/rear) with consistent URLs
Operational guardrails to prevent restart loops when no video arrives
Player UX: don’t drop values to zero when telemetry pauses briefly

Architecture

Built as a practical AWS-native pipeline with reliable operations and predictable playback URLs.

▶

Video ingest + HLS

Dual camera streams are ingested into Kinesis Video Streams and transcoded to HLS, served via CloudFront with stable per-device paths.

≋

Telemetry pipeline

UDP telemetry from the device is ingested via middleware into Kinesis Data Streams, processed by Lambda, and written to S3 as a “latest.json” feed.

⚙

Provisioning + ops

Automated provisioning creates or validates required streams and starts per-device services. A “wait-for-data” runner prevents restart storms when a stream is idle.

Key AWS services

Kinesis Video Streams (dual camera ingest)
CloudFront (HLS distribution)
Kinesis Data Streams (telemetry ingestion)
Lambda (telemetry processing + provisioning triggers)
S3 (latest.json + supporting assets)
SSM (remote provisioning on video EC2)

Player experience

Front stream primary, rear stream PiP
HUD overlay: speed, RPM, throttle, timestamp, and more
Freshness logic prevents brief gaps from showing false zeros
HLS URLs resolved cleanly per camera (front/rear)

Delivery approach

Implemented with clear operational controls and testable components.

What we built

Device provisioning dashboard and backend automation
Per-device video services (front/rear) with canonical HLS playlists
Telemetry middleware integration and cloud processing
HUD web player for live viewing and validation

How we reduced risk

Separated video vs telemetry so failures are diagnosable
Added structured logs and runbook-style checks
Guarded against idle streams to avoid unstable restarts
Validated encoder assumptions via controlled parameter testing

Talk to an engineer Read FAQs

Results

Built for reliability, repeatable onboarding, and real-world field constraints.

Operational stability

Systemd-based per-device services, canonical HLS playlists, and idle-stream handling support consistent uptime behavior.

Scalable onboarding

Provisioning automation ensures new loggers get the right streams, configs, and services with minimal manual setup.

Better viewer UX

The HUD overlay and freshness logic keep telemetry display stable and accurate during transient gaps.

Need a similar real-time system?

If you’re building live video, telemetry/IoT streaming, or multi-stream playback with analytics, we can help you design a reliable pipeline and ship a production-ready implementation.

Book a consultation Explore services Back to Resources