Case Study
Carrier HVAC — Building a Serverless IoT SaaS Platform for Millions of Connected Devices
Carrier HVAC
Role
AWS Solutions Architect & Serverless Application Developer
Period
June 2021 – November 2023
Duration
29 months
Reporting Line
Dual reporting — VP (Technology) and VP (Business)
Industry
Connected HVAC / Smart Buildings
The Situation
Carrier HVAC was investing in a global IoT SaaS platform as a strategic connected-buildings play — sensors and controllers across large facilities monitoring air temperature and air quality at scale.
The customers weren't small: stadiums (Atlanta Braves), enterprise data centers (AWS), and other large-facility operators with sensors per floor and section, generating millions of data points daily that the platform needed to ingest, process, and surface back to operators.
They needed a serverless, event-driven AWS platform built in TypeScript — infrastructure-as-code, APIs, persistence, identity, and cost controls — that could scale from day one without traditional server management overhead.
Why Bring Me In
Carrier needed a senior engineer who would personally design and build the serverless platform — IoT Core ingestion, TypeScript Lambdas, CDK constructs, GraphQL APIs, and polyglot data stores — while guiding frontend and backend teams through the patterns. Not an advisory engagement. A build.
What I Built
- •
Architected the IoT data plane. End-to-end ingestion and processing with AWS IoT Core, Kinesis Data Streams, Lambda, and EventBridge — real-time vs. batch trade-offs, device authentication, message validation, and error handling at airline-facility scale.
- •
Built the serverless application layer in TypeScript. Lambda microservices, Step Functions orchestration, API Gateway, and a GraphQL API layer that cut client API calls by 60% through efficient query design and batch processing.
- •
Designed polyglot persistence for each workload. DynamoDB for high-velocity device state, Timestream for time-series telemetry, Neptune for device relationship graphs, MongoDB for flexible documents — with indexing and access patterns tuned per store.
- •
Established Infrastructure as Code with AWS CDK. TypeScript CDK constructs for repeatable, versioned, testable infrastructure — reusable patterns adopted across multiple services.
- •
Built B2B identity and security. Okta SSO with JWT validation, API security patterns, least-privilege IAM, and encryption at rest and in transit for the multi-tenant platform.
- •
Owned cost engineering and observability. S3 lifecycle policies, DynamoDB capacity strategy, Lambda memory tuning, and CloudWatch log retention delivered 40% infrastructure cost reduction. CloudWatch, Grafana, and CloudTrail dashboards with SLIs/SLOs for operational visibility.
- •
Laid the foundation for future ML. Kinesis → Lambda → S3 ingestion pipelines established the data layer for predictive analytics on building telemetry.
Business Outcomes
- •
99.9% uptime on a global IoT SaaS platform serving 50,000+ B2B users
- •
40% infrastructure cost reduction while supporting millions of daily device events
- •
50% reduction in time-to-market through serverless architecture and IaC automation
- •
Enabled a new IoT SaaS revenue stream for Carrier's connected-HVAC strategy
What I Left Behind
- •
A production IoT SaaS platform serving 50,000+ B2B users at 99.9% uptime
- •
Documented serverless architecture patterns and reusable CDK constructs
- •
GraphQL API layer and polyglot persistence strategy the team could extend
- •
A data foundation ready for future ML and predictive analytics