Toyota.
Intern at Toyota’s software development and autonomous driving group
Autonomous Driving · Vehicle Engineering · Test Engineering · Summer 2025
The Problem.
Poor performance of autonomy features: lots of errors.
The demo-day coming up with execs of Toyota and Woven. Lots of money was invested. Failure was unacceptable.
The Solution.
Root cause analysis and triage of testing issues.
Design components for self-driving test vehicles.
Increase vehicle fleet —> more data.
The Approach.
Designed a sensor enclosure. CAD modeling (SolidWorks).
Ensured compatibility with existing assemblies.
Triaged on-road testing issues daily.
Rode testing vehicles and experienced problems first-hand.
Documented, analyzed, and reported testing issues to software developers and created countermeasures.
The Challenge.
Modeling with complex industrial design surfaces.
Misleading data and sensor signals.
Safety and regulations. Live testing on Tokyo roads.
Comms with Japanese OEMs + Palo Alto engineering teams.
The Fix.
Compared sensor data against simulation and camera views.
Tolerance analysis (tolerance stacks).
Applied surface modeling and DFM/DFA techniques.
Piloted and tested auto-triage: AI-driven testing analysis.
Developed automated, AI-based workflows (Python, Linux).
The Outcome.
Delivered a production-ready enclosure: deployed across 30+ autonomous test vehicles.
Resolved 150+ autonomous driving incident tickets. Improved KMPI (error frequency) by 75%
Trained AI-triage model: increased classification accuracy and implement human validation.
My Contributions.
Enclosure design + attachment method.
Daily on-road testing and data analysis.
AI-triage development and training/validation.