- Best First Year TeamNASA Lunabotics Challenge, 2025
- Runner-upSMTA Bright Electronics Design Challenge, 2026
- International finalistC2C Capture the Flag, 2025
- Three hackathon winsStudent hackathons
Work
Things I have designed and built, from PCB to autonomy stack. Each one ran somewhere outside my desk.
Dermis tactile modules
Modular tactile sensor tiles on magnetic elastomer skin, with firmware that calibrates itself and a data layer that does not care which hardware is underneath.
UB Lunabotics rover
Autonomous excavation rover for NASA's Lunabotics Challenge. Each subsystem runs on its own STM32 node over CAN, under a ten-package ROS2 autonomy stack. Best First Year Team, 2025.
BCI analog front end
Isolated 4-channel biopotential front end with a wireless nRF52840 controller. Power budget and rail topology validated; fNIRS extension planned.
Wrist OS
A custom operating system for an ESP32 watch. The watch listens; the phone transcribes and turns what it hears into notes and tasks. Nothing is stored on the wrist.
All projects 13
- 2026 Dermis tactile modulesModular tactile sensor tiles on magnetic elastomer skin, with firmware that calibrates itself and a data layer that does not care which hardware is underneath. Sensing, Hardware In progress
- 2024 UB Lunabotics roverAutonomous excavation rover for NASA's Lunabotics Challenge. Each subsystem runs on its own STM32 node over CAN, under a ten-package ROS2 autonomy stack. Best First Year Team, 2025. Robotics, Embedded In progress
- 2026 BCI analog front endIsolated 4-channel biopotential front end with a wireless nRF52840 controller. Power budget and rail topology validated; fNIRS extension planned. Hardware, Sensing In progress
- 2026 Wrist OSA custom operating system for an ESP32 watch. The watch listens; the phone transcribes and turns what it hears into notes and tasks. Nothing is stored on the wrist. Embedded, Software In progress
- ResQMeStarted at HackHarvard, now a research platform: a custom LoRa mesh stack on ESP32-C6 and SX1262 with a dual-radio ESP-NOW and LoRa architecture. Embedded, Networking In progress
- 2026 Tactile data gloveBuilt for Cal Hacks: a sensing glove with silicone-cast magnetic fingertip pads and piezo elements for collecting robot training data. Sensing, Hardware Shipped
- 2026 ScopewatchA gateway that checks an AI agent's tool calls before they execute. Built at the Nebius x NVIDIA Global AI Hackathon. Software, AI Shipped
- Home lab voice pipelineESP32-S3 microphone to WebSocket to whisper.cpp to an LLM to Piper TTS, running on a Jetson in a Proxmox home lab. Embedded, Software In progress
- GuardianHATAn ESP32-S3 wearable with sensor fusion and a BLE and Wi-Fi mesh. Embedded, Sensing Shipped
- A-LAD ProAcoustic fault detection: a five-microphone array running TinyML, reporting over ESP-MESH and LoRa. Embedded, Sensing Shipped
- STM32F722 flight controllerA custom flight controller board around the STM32F722. Embedded, Hardware Shipped
- HexaCoreA modular robotics platform built around a Compute Module 5 controller. Robotics Archived
- PinguoA Chinese-learning app with a Django backend and React Native client, deployed on AWS and recognized at an innovation showcase. Software Shipped
Research
Open problems I am working on, where the answer is not known yet.
dermis
Co-founder and CEO
Robots can see. Now they can feel.
Tactile sensing infrastructure for robot hands.
Robot hands can see what they are grabbing but rarely feel it. Dermis builds modular, self-calibrating tactile sensor modules and the software that turns their raw signals into one clean, hardware-independent tactile data stream.
We do not build robot hands. We give the teams who do a reliable way to add touch.
The problem
Tactile sensors today are mostly one-off research builds: hand-calibrated, fragile, and tied to a single gripper. Every manipulation team ends up writing its own drivers, calibration routines and signal processing for each sensor it tries.
What we build
Sensor tiles on magnetic elastomer skin that combine to cover a fingertip or a full palm. Firmware that calibrates itself and reports sensor health. A data layer that outputs the same tactile format whatever hardware sits underneath.
Where we are
Working MVP, shown at Rockwell Automation's expo in Boston in November 2026 for customer discovery. Next: validation on a 7-axis collaborative arm.
Progress
- Picked the problem
Chose a multimodal capture kit for robot training data, built on magnetic elastomer patches we fabricate ourselves.
- First tactile glove
Silicone-cast magnetic fingertip pads with piezo elements, built at Cal Hacks.
- Pivot to a B2B sensing platform
From one glove to modular, self-calibrating sensor tiles and a hardware-independent data layer.
- MVP at Rockwell Automation's expo
First conversations with automation engineers in Boston.
Work with Dermis
Pilots, integrations and research collaborations.
- Website
- dermistouch.com
Failures 3
What broke, why it broke, and what I changed afterward. Most of what I know came from this list.
-
Symptom
The excavation section inherited from our reference design had structural problems we could not tune away.
Root cause
One cantilevered mast carried both bucket drums on a shared rigid arm and shaft, so every load path ran through the same joint.
Fix
Redesigning around a triangulated truss, with each drum on its own linear rail and motor.
What I do differently now
When I borrow a design, I review its load paths before its features.
-
Symptom
ZED camera topics would not stream in ROS2 while we were debugging on the rover.
Root cause
The installed ZED SDK was 5.3.0; the ROS2 wrapper targets SDK 4.x.
Fix
Matched the SDK to the major version the wrapper is built against.
What I do differently now
Vendor SDK and wrapper versions get pinned together in the setup script, and checked before any field session.
-
Symptom
The Jetson Orin Nano fell behind as soon as the ZED depth pipeline and AI inference ran at the same time.
Root cause
Each workload fit on its own; together they exceeded what one module could sustain. The ZED 2i is also USB3, not CSI, which added load the original plan did not account for.
Fix
Onboard compute moves to a mini PC with an RTX 4060-class GPU. Motion control moved to a separate deterministic MCU, so a stalled autonomy computer cannot take the drive system with it.
What I do differently now
I budget compute for the worst concurrent load, not each workload alone, and keep safety-critical control off the machine that does perception.
Trajectory
Roles run in parallel more often than in sequence. The chart shows them that way.
- Now
Tactile sensing infrastructure for robot hands.
- Co-founder & CEODermis
Tactile sensing infrastructure for robot hands.
- Systems Engineering InternAskmae.ai
Founding engineer on an AI-powered wearable for people with disabilities. Details under NDA.
- Co-DirectorUB Hacking
Ran sponsorship and operations for UB Hacking 2026.
- Community AdvisorWilkeson Quad, UB
- Co-founder, Hardware Lead, then Project ManagerUB Lunabotics
Co-founded the team in April 2024. Led computer hardware through our first NASA Lunabotics Challenge (Best First Year Team, 2025). Project Manager for the 2027 team.
- B.S. Computer ScienceUniversity at Buffalo
Graduating May 2027.
Log
Where I have been, what I attended, and what shipped.
-
Took the Dermis MVP to Rockwell Automation's expo
Showed the tactile modules to automation engineers and integrators for validation and customer discovery.
-
Launched Dermis publicly
Dermis is now a B2B tactile sensing platform: modular sensor tiles, calibration software and a hardware-independent data layer.
-
Directed UB Hacking 2026
Two days of building at UB. I led the team and the sponsorship drive.
-
Runner-up, SMTA Bright Electronics Design Challenge
-
Best First Year Team at NASA's Lunabotics Challenge
First year competing as UB Lunabotics, a team I co-founded in April 2024.
Writing
Technical articles, papers and longer stories.
Publications
-
SLAM Doesn't Tell You It's Dying
Looks at ORB-SLAM3's internal signals as early predictors of tracking failure, and reviews three approaches to the problem: IV-SLAM, Matthee et al., and Han et al.
The Embedded EditArticle, Aug 2026
Stories on Medium
Photos
Labs, launches, competition floors and the places in between.
Contact
Let’s build something interesting.
Open to conversations about robotics, embedded systems, AI products, and ambitious technical teams.