Robotics

Senior Platform Software Engineer at Medtronic

June 2023 – Present

Served as the core Software Module Steward for the Windows-based OS infrastructure powering a safety-critical, multi-axis surgical robotics platform. Maintained system integrity, low-latency data distribution, and lifecycle development within a strict FDA-regulated and cybersecurity-focused environment.

Platform Stewardship & Compliance: Headed software lifecycle maintenance and OS configuration management, implementing strict security patches and deterministic controls to meet FDA and medical device cybersecurity standards.
Developer Experience & Tooling Modernization: Standardized development pipelines across multi-disciplinary engineering teams by integrating and vetting AI-assisted development tools to increase code delivery velocity.
Windows Secure Boot CVE Mitigation: Resolved a critical OS-level vulnerability by developing an automated PowerShell installation script for Microsoft security KBs. Engineered a secure multi-reboot persistence protocol to retain system permissions within a highly restricted Windows runtime environment, unlocking future deployment capabilities.
V&V and Field Integration: Packaged, digitally signed, and thoroughly tested the deployment patch. Authored comprehensive Verification & Validation (V&V) protocols and technical work instructions, successfully collaborating with field service engineers to iterate on UI/UX preferences—such as integrating QR codes to replace raw CLI commands.
Distributed Systems & Cryptographic Optimization: Supported distributed network debugging by collaborating directly with middleware domain experts to isolate asynchronous C++ messaging bottlenecks within an RTI DDS (Data Distribution Service) framework. During a core DDS version upgrade and cryptographic library migration, solo-diagnosed a critical latency spike caused by missing optimization flags in the compilation build; rebuilt and deployed the corrected binary to slash maximum communication latency from 20ms down to 5ms.

Beachbots Major Qualifying Project

August 2019 – May 2020

Co-designed and built a multi-robot beach cleaning system designed for autonomous task scheduling and waste retrieval. Owned the physical hardware-software integration and co-authored the high-level control stack using ROS and Python.

Hardware-in-the-Loop Integration: Integrated a heterogeneous component stack including a stepper motor for collector shoulder articulation, a precision servo for arm/gripper manipulation, physical buttons for hardware calibration, and wheel encoders for closed-loop drivetrain telemetry.
Kinematics & Control: Programmed the Python manipulation module, translating analytical inverse kinematics equations from the mechanical team into deterministic joint-space motor profiles.
State Machine Design: Co-developed the central ROS-based finite state machine to manage task sequencing and multi-agent coordination.
Constraint Management & Simulation: Navigated physical drivetrain failures in sandy terrain and severe budget limits by pivoting to a high-fidelity virtual proof-of-concept. Developed a simulation environment alongside a co-developer to validate the system architecture, achieving 100% successful state transitions when the "SmallBot" received instructions from the supervisory "BaseBot."

click here to view the report

Robotic Arm Ball Sorter Project

January 2019 – March 2019

Developed a 3 degrees-of-freedom (DoF) robotic manipulator utilizing computer vision and advanced kinematics to autonomously sort spherical objects based on size and color metrics.

Mathematical Modeling & Kinematics: Derived and implemented the forward kinematics, analytical inverse kinematics, and Jacobian matrices to govern precise spatial movements and avoid singularity states during manipulation tasks.
Trajectory Generation & Simulation: Translated geometric control equations into executable Python and MATLAB trajectory generation scripts, optimizing joint velocities and smooth path execution.
Telemetry Visualization: Engineered a real-time MATLAB data visualization suite to plot live joint positions and end-effector coordinate spaces, enabling instant verification of tracking accuracy against target paths.
Team Collaboration & Result: Collaborated within a three-person cross-functional team to integrate the perception and control pipelines, validating the mathematical model through successful physical object classification and sorting cycles.
Diagnostic Triage & Signal Analysis: Leveraged the custom visualization tool to identify a system anomaly regarding sporadic differential kinematics calculation vectors (J ⋅ q̇). Diagnosed high-magnitude velocity vectors during zero-motion phases by auditing numerical differentiation scripts and cross-referencing telemetry directly from an STM32 Nucleo firmware status server to isolate the noise profile.

click here to view the report

Maze Mapping Turtlebot Project

October 2018 – December 2018

Co-developed an autonomous mobile robot navigation stack in Python using ROS to map unknown environments via SLAM and execute point-and-click goal navigation within the generated grid map.

SLAM & Mapping Infrastructure: Configured and deployed the ROS gmapping package to execute Simultaneous Localization and Mapping (SLAM), generating dynamic 2D occupancy grid maps using live LiDAR telemetry.
Frontier Exploration & Task Allocation: Built a Python script integrating frontier finding and clustering algorithms to automatically detect boundaries between known and unknown spaces to drive exploration.
Path Planning & Navigation Stack: Authored a custom A* pathfinding algorithm in Python to calculate efficient, collision-free global paths from the robot's current pose to a user-selected coordinate goal.
High-Level Behavior Control: Developed a supervisory ROS node and motor driver interface to translate planned coordinate paths into deterministic velocity commands (geometry_msgs/Twist).
Diagnostic Triage & Constraint Management: Resolved a physical hardware flaw where protruding wheels hit obstacles by programmatically expanding the SLAM map wall inflation buffers. Overcame sensor limitations by filtering out transient, single-cell noise from the LiDAR stream to prevent erratic global path calculations.

click here to view the presentation