OBDPulse is a commercial vehicle diagnostics platform that turns any smartphone into a professional-grade OBD-II scanner. Built at Pushkal Technologies as part of a 2-developer engineering team, I owned the full technical stack — mobile app, serverless backend, admin dashboard, and production deployment.
Mobile App (Flutter) Built a cross-platform Flutter app shipping on both Android and iOS, supporting real-time vehicle diagnostics over Bluetooth Low Energy. The app reads and clears fault codes, streams live sensor data (RPM, speed, VIN, freeze frame, O2 sensors), runs service resets, and includes a Track Mode for performance driving. Shipped with full localization across 17 languages including Arabic, Japanese, Korean, Chinese, Russian, and more.
One of the trickiest engineering problems was taming BLE's stream-based communication model. OBD-II devices communicate via raw read/write streams with no native request-response pairing. I designed a queue-based abstraction layer that converted the stream into a clean async function pattern ——— input → await → output ——— with proper ordering guarantees. Additionally, at high polling rates the OBD device occasionally emitted duplicate telemetry frames; solved this by implementing a debounce filter on the data stream before it reached the UI layer.
Backend (Go + AWS Lambda + DynamoDB) Designed and implemented a serverless microservice architecture using Go on AWS Lambda. Chose Lambda over EC2 deliberately — the workload (user sessions, purchase verification, scan logs, analytics events) was event-driven and sporadic, making Lambda's cost model and free tier ideal. DynamoDB was a natural fit for its native AWS integration and schema flexibility across device telemetry types. The backend handles user auth (Google, Apple), in-app purchase validation for both platforms, scan logging, and analytics aggregation.
Admin Dashboard Built an internal admin dashboard for monitoring platform health — real-time user analytics, purchase tracking across iOS and Android, reported issue management, and scan volume trends.
Production Impact
- 2,700+ registered users across Android and iOS
- 3,900+ total vehicle scans performed
- 60 paying subscribers (Plus and Pro tiers)
- 17-language localization covering major global markets