Ajay Suresh

Argonne National Laboratory

Led battery systems development and software validation for a DOE program where 12 university teams of students designed, built, and tested EV battery packs within production vehicles. Now building AI-native systems and operational tooling alongside battery and vehicle systems R&D (projected to save $80k / year in person-hour costs).

In the Battery Workforce Challenge, teams were building and testing their first high-voltage battery packs. There was no shared infrastructure to prove that a student-developed BMS was safe to operate on real cells, and twelve teams inventing their own validation frameworks would have meant twelve different definitions of safe.

So validation became infrastructure. A staged framework took each team from bench-level tests through model- and hardware-in-the-loop validation to full-pack end-of-line and in-vehicle tests within just a few months. The test platforms were deliberately low-cost and repeatable (MATLAB/Simulink, CAN tooling, Python) to allow for team-defined flexibility with program-enforced rigor.

Teams built full battery packs with production-intent; the strongest ones completed validation across model, hardware, and pack level and went on to vehicle integration and calibration. Despite the accelerated timeline, we progressed multiple teams through pack-level testing, and impressed all industry partners including Stellantis’ CEO.

The frameworks we developed are now described in two IEEE ITEC+EATS 2026 papers, currently pending publication.

Photos: AVTC.

UC Davis EcoCAR

Built a team, program and organizational infrastructure from scratch, recruited and managed junior engineers, and executed the design, testing, and integration of a custom powertrain and autonomous stack into a production Cadillac LYRIQ.

UC Davis had been out of the national Advanced Vehicle Technology Competition series for fifteen years, and while I had industry experience, I had zero experience with vehicles before starting. The program had to exist before it could deliver anything: recruiting, garage and workspace setup; acquiring automotive-standard tooling, developing a safety and information-security regime, and integrating the digital workflows that coordinated complex vehicle development tasks amongst dozens of students.

The engineering work ran on an adopted V-model. We embraced a detail-oriented approach to systems modeling, component, subsystems, and systems-level xIL testing, and a test-first approach even under a compressed timeline.

Our team scripted an undeniable turnaround on the national stage, going from placing 11th out of 13 teams in the first year to placing 5th overall and bagging over 8 awards across domains in Year 2. We also were one of the first few teams to get a vehicle working and on the road (video below)!

Tesla

Product engineering on the Powerwall fleet: customer cases, failure analysis on returned units, install and commissioning quality, and telemetry-driven root cause analysis.

A battery pack spends a majority of its life in the field, and problems arrive one support case at a time. Diagnosing each case in isolation does not scale, and it wastes what the fleet already knows: every unit reports the same telemetry, so one confirmed fault is a pattern you can go look for everywhere.

The working loop: starting from a single case and its data, forming a hypothesis, expressing it as a query the whole fleet can answer, correlating the hits against firmware versions, environmental factors, and usage, confirming against physical failure analysis on returned units, and leaving behind an automated containment system so the next occurrence is caught before a customer notices. Around that loop sat the physical half of the job: inspecting returned units, install and commissioning quality, and on-site debugging. I repeated this pattern across hundreds of customer cases, detecting and solving hardware, software, and manufacturing quality issues. Just within my first year, I brought our SLA for actioning support-team escalations from months to days, and coordinated countermeasures for the top 70% of fleet-wide hardware failures.

I also returned to Tesla while completing my master's degree to work the same problem from the design reliability perspective: test fixtures, automation, and the data and modeling pipelines that reliability decisions depended on. Both times, what the field taught me was key to informing future iterations of products.

Solarlytics

First role out of school, at a stealth solar energy startup. Built a self-powered panel-instrumentation system end to end, from silicon to cloud.

At a stealth startup where every engineer wore multiple hats, the instrumentation system had one owner, me (a new grad with no experience): working on circuitry and PCB, register-level firmware for our microcontrollers, secure connectivity, OTA updates, data infrastructure, and a full stack application that controlled our instruments and turned raw telemetry samples into waveforms someone could actually look at.

It is also where I learned that an instrument is only half the product. The value showed up when the signal left the device: high-frequency measurements published reliably enough to reconstruct remotely, and test automation that turned hours of board bring-up into minutes.