Full Stack Engineer - Traffic Studies

About the role

ABOUT OBVIO AI

Each year, more than 40,000 people in the U.S. leave home and never make it back due to traffic crashes. At Obvio, we believe these deaths are preventable. We deploy solar-powered, AI-assisted cameras to enforce traffic laws where pedestrians are most vulnerable, automating enforcement in ways that traditional systems cannot. Our approach has already led to a 50% reduction in reckless driving in early partner cities.

Founded by the team behind Motive's AI dashcam and backed by Bain Capital Ventures and Khosla Ventures, we are building the intelligence layer for safer streets globally.

ABOUT THE ROLE

Obvio's traffic studies team runs short, camera- and radar-based data studies that measure speeding, red-light, stop-sign and crosswalk behavior at specific locations in a city before any enforcement program starts. Cities rely on these results, so accuracy and timely delivery matter.

Today the pipeline works, but it scales with people rather than software. We are hiring a Full-Stack Engineer to change that. You will own the software that takes a study from field deployment to a published report, make the routine path repeatable and largely automated, and make ad-hoc requests (a new city, a new sensor, a new metric, a tighter timeline) something the system takes on through configuration and small extensions rather than one-off engineering effort.

You will work alongside computer-vision engineers who hand you detectors and trackers, an annotation team that reviews and audits the results, field operations who deploy the hardware, and a go-to-market team that sets delivery dates with cities. Your job is the connective tissue between them.

WHAT YOU'LL DO

- Own the traffic-study backend end to end: ingestion from multiple sensor types, vision-pipeline orchestration on our ML servers and cloud compute, event extraction, annotation hand-off, statistics, and publishing to the city-facing portal (e.g., Python services on AWS ECS/EC2, S3 event data, Postgres).

- Scale the routine path: turn manual steps (stats entry, reprocessing, report assembly) into automated, auditable ones so a study runs the same way every time with minimal engineer involvement.

- Handle ad-hoc requests: new sensors (each with its own data layout and quirks), new metrics, new report formats and shorter timelines. Design the system so these become configuration and small extensions rather than rewrites.

- Design and evolve the data model: schema design, indexing, and query performance on growing Postgres datasets that back both the pipeline and the internal tools built on top of it.

- Own pipeline health: maintain monitoring and alerting for the vision and event-extraction services, catch stuck or failed studies early, and drive fixes to root cause.

- Make the numbers defensible: implement counting and violation logic from written specifications, version it, and make every reported figure traceable to the raw data and the logic that produced it.