Work Experience

Building production robotics systems that deliver real-world impact

Software Engineer

OSAAP

Chemsford, Massachusetts

Apr 2026 - Present

  • Integrated a Canon EOS camera into OSAAP's shadow board manufacturing pipeline via a custom C++ wrapper, building in-browser image quality validation checks that give operators immediate capture feedback and reduce downstream failures
  • Developing OpenCV-based perception features for OSAAP's tooling to reduce operator intervention in the scanning workflow
  • Integrated a Canon EOS camera into OSAAP's shadow board manufacturing pipeline via a C++ EDSDK wrapper, enabling reliable hardware-triggered image capture with configurable exposure and framing parameters aligned to BlueShadow scan requirements.
  • Extended the camera station perception stack with image preprocessing and EXIF metadata extraction, ensuring captured tool images reach the Azure-hosted BlueMobile portal with consistent quality for reliable downstream processing.
ROS C++ Python Path Planning Perception CI/CD Docker

Robotics Software Engineer II

Vecna Robotics

Waltham, Massachusetts

Mar 2024 - Oct 2025

  • Designed and implemented task planning and congestion management modules with ROS actionLib, costmaps, and priority scheduling, reducing multi-robot contention by ~25% and increasing throughput in busy warehouse zones
  • Refactored the sensor driver stack (Waymo/Hesai LiDARs; Basler/Lucid cameras), yielding ~15% acceleration in new forklift platform development with backward compatibility intact
  • Enhanced pallet detection by standardizing and expanding the pallet detection library to include 25% more types, which increased pickup success to 95% from around 82%, and enabled deployment in a wider range of compact site scenarios
  • Delivered customer-driven autonomy features by implementing adaptive, height-aware pallet docking and integrations for site-specific equipment, unlocking new customer workflows and increasing per-site expansion potential by up to 40%
  • Spearheaded on-site development and validation of high-risk, time-constrained features, cutting facility integration cycles by ~30% and support interventions by 8-20%, resulting in increased customer trust and expansion potential
  • Led contributions across the autonomy stack (navigation, perception, task allocation, path planning, recovery strategies) and built dependable CI/CD and containerized delivery, improving overall performance on owned sites by 4-15%
  • Collaborated with hardware and firmware teams to incorporate up to 14% of custom features for new sensing modalities (IR). Developed state machines that integrate with the existing stack, increasing support for high-risk tasks by 10%
ROS C++ Python Path Planning Perception CI/CD Docker

Robotics Software Engineer I

Vecna Robotics

Waltham, Massachusetts

Jan 2022 - Mar 2024

  • Contributed to end-to-end development, testing, and deployment of the autonomy stack for heavy material-handling robots, increasing system reliability and cutting unexpected robot stoppages by ~22% across customer sites
  • Improved pallet and obstacle detection modules by applying point-cloud clustering and upgrading filtering pipelines, improving data utilization and raising detection accuracy by 6-10%
  • Developed tools to analyze new feature impacts from prototypes; collaborated with deployment to improve testing, optimizing rollout by ~18% and boosting customer adoption
  • Resolved critical customer field issues via ROS bag analysis, hotfix deployments, and autonomy-stack patching, reducing average incident resolution time from more than 5 days to less than 48 hours
  • Integrated autonomy features on high-reach platforms, such as pallet drop and reach state-machine management, improve drop efficiency by around 20% and ensure safety in multi-level warehouses
  • Refined internal tools to improve Vecna remote support, raising mission success from ~85% to 96%
ROS Python C++ Computer Vision Point Clouds Debugging

Computer Vision Engineer Intern

Vidalign Inc.

College Park, Maryland

Jul 2021 - Dec 2021

  • Designed monocular 3D reconstruction pipelines, increasing data throughput by ~35% and reducing GPU hours by ~29%
  • Enhanced facial landmark detection & tracking with temporal filtering/outlier rejection, improving accuracy (NME 3.7% to 2.9%) and robustness while sustaining real-time rates (~60 FPS)
  • Optimized integration between reconstruction and tracking, reducing latency 120ms to 74ms (−38%), increasing mesh FPS 18 to 29, and lowering registration failures ~41%
Python OpenCV 3D Reconstruction Computer Vision Real-time Processing