Led production deployment of core perception systems for safety-critical industrial autonomous vehicles, spanning LiDAR ground segmentation, dynamic occupancy representations, and fail-operational diagnostics.

Improved real-world object detection performance and enabled smoother, more reliable trajectory generation in dense industrial ODDs.

Built the evaluation and verification infrastructure required to ship safely, including a metrics framework and simulation-driven V&V across Model-in-the-Loop (MIL), Software-in-the-Loop (SIL), and Hardware-in-the-Loop (HIL) testing, enabling early detection of regressions and real-world failure modes.

Owned production readiness end-to-end, from requirements coverage and simulation-based validation to long-term maintainability, reducing technical debt and increasing unit test coverage.

Fully autonomous urban vehicles (Origin platform)

Worked on integration and production deployment of autonomy software for the Cruise Origin, a purpose-built, driverless vehicle operating in dense urban environments.

Owned end-to-end integration of the autonomy stack with vehicle hardware, spanning simulation, bench testing, closed-course validation, and limited public-road operation. This included bringing up new releases on real vehicles, diagnosing cross-stack failures, and resolving hardware–software coupling issues that only emerge at system scale.

I also managed and optimized the production launch pipeline for autonomy releases, enabling repeatable, safe bring-up of perception and autonomy components under constrained, safety-critical operating modes.

*Impacted by company shutdown in December 2022.

Worked on safety-critical autonomy systems focused on redundancy and fail-operational behavior for autonomous vehicles.

Designed and optimized components of a redundant AV stack intended to safely take over when the primary autonomy system degrades or fails, with emphasis on perception robustness and predictable system behavior under fault conditions.

Integrated high-density custom LiDAR data (~900k points per sweep) into ground surface estimation pipelines, improving perception accuracy while reducing average runtime from 42 ms to 18 ms to meet real-time constraints. Built internal tooling to visualize and debug ground estimation outputs and failure cases directly from raw LiDAR streams.

This work reinforced the importance of designing autonomy systems that degrade gracefully, expose failure early, and remain predictable under partial system loss.

FDA-regulated robotic surgical platforms

Spent seven years building and scaling FDA-regulated, safety-critical robotic systems used in live clinical environments, where design errors translate directly to patient risk, recalls, or production shutdowns.

Led mechanical design and production validation for next-generation advanced energy surgical systems, owning end-to-end development from early architecture through design verification, validation, and manufacturing scale-up. This included thermal management of power electronics, durability and safety testing (shock, vibration, thermal, fluid ingress), and close vendor collaboration to ensure DfMA and production robustness.

In parallel, worked on sustaining engineering and supply-chain resilience for advanced energy instruments in active clinical production, including COVID-era supply disruptions. Built internal quality analytics to identify systemic failure modes, prioritize high-risk components, and drive continuous quality and cost improvements across multiple product lines.

This experience shaped my approach to building physical systems: assume failure, validate aggressively, and design for scale and long-term reliability, not just first launch.

▪ Developed a Human Gait State Estimator for Prostheses based on an EK filter which was able to successfully estimate the gait state with 98.5% accuracy of 10 test subjects.