Anduril application · Vehicle software · Costa Mesa
Operator interfaces for software and physical systems.
I have built rover software, robotic controls, and operational applications. My work includes Rust interfaces to embedded controllers, React/TypeScript operator tools, and Python services. I have debugged hardware, tested changes through simulation and replay, and trained the people using the finished system. I want to apply that experience to Anduril’s vehicle software.
Saddleback College Robotics
My work connected autonomous navigation, manual controls and the interfaces a multidisciplinary team used to operate the rover.
ROS 2 · Rust · STM32 · CAN · Operator interfaces · Field testing
I built React/TypeScript and Foxglove interfaces for rover autonomy and manual control, supporting engineers, chemists, and physicists conducting science tests. The interfaces presented vehicle telemetry, waypoint commands, and Jetson video over a radio link.
I connected operator commands to STM32 through an asynchronous Rust driver and ODrive CAN controls, then investigated failures during field tests and rosbag replay.
Interactive simulation built from repository arm geometry. The route, sample, fixtures, and proposed tool-camera view are simulated. The arm geometry comes from the repository; no recorded camera feed is shown.
I co-led ROS 2 autonomy through two competition seasons and worked on interfaces for autonomous and manual operation. That work connected navigation and vehicle state to the controls available to the operator, while supporting teammates running science tests.
An asynchronous Rust driver connected the ROS 2 system to STM32 controls. ODrive motor controllers communicated over CAN. I worked across the rover software and operator tools to investigate control behavior and coordinate changes with the team.
I tested transitions between autonomous and manual control, radio loss and sensor failures during field trials. Recorded data supported investigation after a run. The focus was whether commands and failure handling produced the expected vehicle behavior.
Ibrium Studios
ARM Cortex-M · CAN · Altium · Logic-analyzer debugging · AWS · Operating procedures
At Ibrium Studios, I delivered a multi-axis robotic control system for a Disney production, starting with producers’ requirements and carrying it through PCB design, board bring-up, firmware integration, and deployment. I designed multilayer boards in Altium and implemented ARM Cortex-M firmware, connecting the nodes to a central controller over CAN. The system included feedback, fault detection, and safe-stop behavior. These controls made failure handling part of the implementation, alongside the checks needed before production use.
When motion became intermittently late, I used a logic analyzer to trace the timing across the control path. The cause was a control-loop deadline overrun. I moved scheduling to a fixed-rate timer interrupt and revalidated the motion sequence. I checked the corrected behavior against the motion sequence before production handoff. Timing traces provided evidence for the diagnosis and a way to verify the fix.
Before handoff, I verified motion sequences for production use and prepared operating procedures and crew training. Delivery meant getting the hardware and software working together under a fixed deadline and making the system usable by the production team. For Depot, I built a shared asset-state model covering ownership, stage, and activity. Active-session conflict detection helped teams avoid interfering with work in progress. Deletion was limited to assets established as unowned or redundant, and AWS shutdown was guarded by activity checks. I remained responsible until the crews could operate the system. I would bring that same involvement to vehicle integration and deployment support.
O3 Worldwide
Rust · Python/FastAPI · PostgreSQL · React/TypeScript · OPC-UA · OR-Tools · Docker
Working with shipyard planners at O3, I designed cloud-based coordination across multiple shipyards. A shared ontology connected schedules, orders, parts, and production progress so teams could coordinate interdependent work and trace updates across sites. I built the application with a React/TypeScript interface, Rust and Python services, and PostgreSQL, connecting the operational model to OR-Tools scheduling.
Late deliveries and unavailable equipment required changes to the underlying dependency model. I extended the logic and checked the revised behavior through historical replay. OPC-UA adapters connected operational inputs; access controls and event history made changes traceable. The software was developed for potential deployment by U.S. shipbuilders, with local deployment, role-based access, and tamper-evident audit records. My broader O3 work includes engineering and planning for the O3-led, $15 billion-plus Orlando Sanford International Airport development opportunity. I also support pilot and technician training, aircraft manufacturing, and digital-engineering work spanning blended-wing-body aircraft, unmanned aerial and surface vehicles, and advanced boats and small ships.
I separated the operational model from the supporting services and interface to support expansion across sites. Shared services and a common data model let the architecture accommodate coordination between shipyards while keeping operational changes distinct from interface changes. My earlier work at MedBWS also involved supporting software after deployment. I built pharmaceutical identification and disposal workflows in HIPAA-compliant hospital environments, with secure APIs and audit logging to control access and make application activity traceable.
Engineering projects
How I processed sensor data, evaluated physical systems, and built tools for engineering review.
Haskell · Go · TypeScript · Nix
I built a counter-UAS simulation and replay system that carries signal detections through to constrained response selection. The Haskell core uses FFT and CA-CFAR detection, extracts emitter features, and matches them against a threat library. Response plans account for mission state, spectrum restrictions, and available resources, with confidence-based abstention when the evidence does not support a decision.
Go services and a TypeScript viewer make the processing and decisions inspectable. Synthetic and recorded IQ inputs, saved configuration, and decision evidence let me reproduce failures and compare changes. The work combines sensor-data processing with operator-facing software and repeatable testing of mission logic.
Python · NumPy · SciPy · OpenVSP · CalculiX
I built a workflow connecting parametric aircraft geometry to aerodynamic, structural, mass, and flutter analysis. It screened hundreds of design variants and advanced three finalists to higher-fidelity analysis. One trade-study design showed approximately 24% lower structural mass than its baseline, with the result retained as an analysis finding for engineering review.
The engineering challenge was maintaining consistency as geometry and analysis outputs moved between tools. I checked those handoffs, retained solver artifacts, compared outputs with published references, and flagged uncertain results. Working with aerospace engineers required making the evidence behind a design decision clear enough to review and reproduce.
Rust · C/C++ · PX4 · MAVLink
I built a Rust static analyzer for PX4 C/C++ flight software to inspect how functions and MAVLink handlers connect. It extracts calls and checks for authentication references, unsafe memory access, and unexpected navigation-state writes, helping focus source review on code that can affect vehicle behavior.
I tested the extraction and detection rules against PX4 function lookups and vulnerable-versus-corrected fixtures. The analyzer passed 29 tests and a release build. Findings remain traceable to source, and reproducible fixtures help distinguish analyzer errors from code that warrants investigation. This work strengthened my ability to examine an existing vehicle codebase and build tools for repeatable software verification.
Hasan & Associates · Computer vision · Spatial analysis
At Hasan & Associates, I combined footage from three body cameras with audio transcripts to reconstruct a scene and examine what each camera could see. I integrated open-source models with our own computer-vision components and validated measurements so the reconstruction could support review of spatial relationships alongside recorded events.
The work required translating a litigation team’s questions into a usable technical workflow. I also built tools for retrieving video, images, documents, and timelines, and delivered courtroom-ready multimedia exhibits. It connects computer vision and spatial modeling with the same user-facing engineering judgment needed to turn complex sensor information into something people can examine and act on.
Technical background
Software development
C++, Rust, Go, Python, TypeScript, React, FastAPI, PostgreSQL
Robotics and testing
ROS 2, Nav2, ARM Cortex-M, STM32, CAN, Gazebo, Isaac Sim, SITL/HITL, rosbag, MATLAB/Simulink
Services and development tools
Linux, AWS, Docker, Git, Nix, WebSocket, Protocol Buffers
My projects include scheduling algorithms, operational data models, asynchronous device interfaces, and repeatable tests. The examples above describe where I used them and how I investigated failures.
Assessments
CodeSignal Industry Coding Assessment
600/600 · August 2026
AFQT Predictor Test
99th percentile · April 2026
Why Anduril
Anduril is where I want to build my career. Learning from mentors at Anduril has strengthened that ambition and shaped the kind of engineer I want to become. I believe in the mission of strengthening American and allied defense, and I want to contribute to it through work I can take responsibility for. I’m looking for the opportunity to keep learning from people I respect while earning their trust through what I build.
I’m based in Irvine and available for full-time onsite work in Costa Mesa and travel. I want to contribute useful software while growing toward ownership of larger vehicle and mission capabilities.
Software Engineer — Undersea Dominance ↗