Research Engineer
Brent A. Thorne
Bridging physical hardware constraints with high-performance scientific software.
With over two decades across embedded systems and computational modeling, I engineer solutions at the full hardware–software boundary — from oscilloscope bench bring-up and Linux device drivers to GPU-accelerated parallel algorithms for reproducible scientific simulation. I build deterministic infrastructure that lets domain scientists and hardware engineers trust their data.
Highlights
Selected achievements
Immediate credibility — commercial scale, scientific tooling, imaging systems, IP, and graduate research. Full detail is in the resume.
- Shipped commercial firmware deployed across millions of consumer and utility endpoints (OpenTV, GridNet).
- Built an execution-agnostic HPC validation framework (hpc-runner) for molecular dynamics workloads across GROMACS, LAMMPS, and OpenMM.
- Developed CT and radio imaging systems with custom reconstruction algorithms and automated artifact correction.
- Inventor on US Patent Application 61834836 — closed-loop biometric monitoring and adaptive physiological feedback.
- Graduate research in scientific computing and parallel algorithms (CUDA, MPI, OpenMP, UPC++) at UC Berkeley MSSE / CS267.
Trajectory
How the work connects
A progression across eras — not a grab bag of unrelated interests. Each step compounds the next.
Embedded Systems
Early career
Wireless Communications
OpenTV → GridNet
Sensing & Measurement
Field & lab
Imaging Systems
Freelance R&D
Scientific Computing
UC Berkeley MSSE
AI-Assisted Engineering
Present
Core engineering domains
Where to go next
Four pathways aligned to the work I do. Each card jumps to that domain’s projects. Stack lines are secondary — problems and outcomes lead.
High-Performance Computing & Scientific Software
Reproducibility and parallel correctness across heterogeneous backends are the bottleneck. Automated regression, benchmarking, and parallel algorithm work for numerical workloads on local, container, and cluster environments.
C++ · CUDA · MPI / OpenMP · Slurm · Docker · GROMACS / LAMMPS / OpenMM
Featured: hpc-runner
Embedded Firmware & Sensing Systems
Board bring-up, hardware-in-the-loop lab automation, and high-rate sensor acquisition — firmware and drivers that survive the path from bench to commercial utility and telemetry endpoints.
Embedded Linux · Device Drivers · DMA · V4L2 · GStreamer · OpenCV
Featured: rad-lab · CEDE / cede-labgrid
Computational Imaging & Instrumentation R&D
End-to-end optical and RF instrument control: bridge raw multi-channel measurements to GPU-accelerated reconstruction and artifact correction. Public proof includes fluorescence spectrometer DAQ bring-up.
CUDA · Signal Processing · OpenCV · DAQ · RS-232 / Serial
Featured: spectrometer-lab · CT / Radio Reconstruction
Scientific Field & Lab Infrastructure
Fixtures, telemetry, and lab automation that isolate physical ambiguity from software behavior — remote observatory and sensor context, RF link discipline, and bioelectrical IP. Historical bench craft lives in the Hardware Archive.
Telemetry · RF Link Budgets · CAD/CAM · CNC · AC/DC Power
Featured: Observatory / telemetry · Bioelectrical IP
Featured
Flagship public work
Constraint-first systems that define the Research Engineer story: HPC regression (hpc-runner), RF HWIL (rad-lab), optical DAQ (spectrometer-lab), and FPGA bring-up continuity (FPGA Kit). Pick a domain above to see the rest of the shelf.
- 01
HPC Runner
PublicPython · HPC · Testing · CUDA
Guarantee reproducibility across backends: an execution-agnostic regression and benchmarking engine for molecular dynamics workloads (GROMACS, LAMMPS, OpenMM) on local CUDA, containers, and Slurm clusters — without rewriting the application for each target.
- 02
rad-lab
PublicPython · SDR · HWIL · RF
RF and sensing bring-up you can evidence: containerized SDR and multi-MCU HWIL with human assent before TX, SIL→HWIL stages, and structured reports — host-gated lab automation, not an ad-hoc bench script.
- 03
spectrometer-lab
PublicPython · Instrumentation · Spectroscopy · RS-232
Bridge optical hardware to quantitative spectra: RS-232 DAQ and control for a PerkinElmer LS 55. Stages 1–3 PASS (2026-10-28) for electrical bring-up, protocol validation, and empty-compartment emission scans; Gate D optical validation remains open.
- 04
FPGA Kit
PublicPython · FPGA · Bring-up · Lab
Keep firmware work alive when toolchains churn: a bring-up archive for Digilent Spartan boards — toolchain memory, board notes, and lab continuity so board work stays portable.
High-Performance Computing & Scientific Software
Clusters, numerics, and reproducible environments
Public compute tooling that makes scientific software portable — schedulers you can recreate, runnable molecular examples, and teaching environments. Flagship: hpc-runner (featured above).
- 01
sci-slurm
PublicDockerfile · HPC · Docker · Teaching
Bring a scheduler stack without a bare-metal cluster — a multi-container Slurm image for development and teaching.
- 02
molecularSim
PublicC++ · Molecular simulation · C++
Make textbook numerical methods runnable — worked C++ examples from Molecular Simulation of Fluids by Richard J. Sadus.
- 03
MESA Docker
PublicShell · Astrophysics · Docker
Remote stellar-evolution work without fighting the host — Docker image for MESA with SSH and pgstar, paired with ASTR 400 setup guides.
- 04
ASTR 400 setup guide
PublicPython · Courseware · Reproducibility
Get students to a working scientific environment fast — practical Python, Git, Docker, and VM setup for SFSU Astrophysics.
Embedded Firmware & Sensing Systems
Drivers, HIL labs, and board bring-up
Featured rad-lab and FPGA Kit carry HWIL and FPGA continuity. Below: containerized multi-MCU labs, kernel drivers, and netboot infrastructure that keep the bench reachable.
- 01
CEDE / cede-labgrid
PublicPython · Embedded · HWIL · Labgrid
Reproducible multi-MCU bring-up that does not depend on one laptop — containerized embedded lab with a Labgrid hardware-in-the-loop port.
- 02
Radar Module
PublicShell / C · Kernel · Yocto · Drivers
Put a radar module under Linux control — kernel driver with Docker image and Yocto environment for BeagleBone-class targets.
- 03
tftp-boot-docker
PublicShell · Infra · PXE · Docker
Lab netboot without ceremony — Docker TFTP and HTTP for UEFI PXE boot of Ubuntu Desktop LTS with OpenWrt DHCP.
Computational Imaging & Instrumentation
From measurements to quantitative insight
Featured spectrometer-lab covers optical DAQ bring-up (Stages 1–3 PASS; Gate D open). Freelance CT/radio reconstruction and artifact correction live in the resume. Public sensing reference below.
- 01
Biofeedback EMG
PublicJava · Sensing · Arduino
Closed-loop physiological feedback from the body — early EMG acquisition still used as a sensing reference for adaptive control ideas.
Hardware Archive
R&D bench & studio — secondary to the Research Engineer story
Historical retro-computing, ferrite planes, and CRT drive work show craft and curiosity. They support the identity above; they do not redefine it.
- 01
Core Memory Redux
PublicPython · Hardware · Simulation · SIL
Prove the driver in simulation before physical risk — fidelity-gated SIL ladder for a 64×64 coincident-current core-memory plane, from SPICE through ideal N×N models.
- 02
Core Memory
PublicPython · Hardware · Electronics
Bring a real 9″ ferrite plane back under software control — drive and sense a 64×64 array (4,096 bits).
- 03
CRT Drive
PublicC · Firmware · Retrocompute · Hardware
Replace sealed drive ASICs on the bench — open firmware that generates sync and video for analog CRTs and TVs from a microcontroller.
- 04
Retro Compute
PrivatePython · Retrocompute · Study
Study guide for Yaohan Chu, Digital Computer Design Fundamentals (McGraw-Hill, 1962) — still private until ready to publish.
Private for now — will link here if published.
Not published
Experience
Professional history
Industry depth across embedded firmware, wireless systems, and independent R&D. Reviewers care about this before degrees — full detail is in the resume.
Dec 2018 – Present
Research Engineer / Scientific & Systems Consultant
Freelance / Independent R&D · San Francisco, CA
- Build computational systems that bridge physical instrumentation, embedded hardware, imaging modalities, and high-performance software.
- GPU-accelerated C++/CUDA and Python pipelines for CT/radio reconstruction and artifact correction; microwave resonant substrate analysis; ECAD/PCB and MCAD fixtures for firmware validation.
- Containerized SDR and multi-MCU HWIL labs (rad-lab, CEDE/cede-labgrid), fluorescence spectrometer acquisition (spectrometer-lab), and high-rate sensor pipelines for synchronized telemetry.
Oct 2013 – Nov 2018
Senior HW/SW Research & Development Engineer
GridNet · San Francisco, CA
- Embedded radio modem firmware and Linux kernel networking drivers for sub-GHz smart-grid telemetry and LTE CAT-4/CAT-M M2M platforms.
- SDR protocol analysis and cross-layer debugging across custom RF hardware, firmware, and embedded Linux runtimes in high-interference environments.
2012 – Present (Concurrent)
Instructor / Educator & Community Mentor
TechShop & Humanmade · San Francisco, CA
- Hands-on workshops in CNC machining, digital fabrication, microcontroller programming, CAD/CAM, and shop safety for research engineers and makers.
- Guided entrepreneurs and researchers through rapid prototyping, board bring-up, and electro-mechanical assembly.
May 2011 – Oct 2013
Computer Scientist, Design Engineer & Inventor
Freelance Consultant · San Francisco, CA
- Mechatronic product development spanning CAD/CAM, CNC, PCB layout, and embedded software across multidisciplinary teams.
- Bioelectrical acquisition front-ends and wavelet analysis for real-time physiological feedback (US Patent Application 61834836).
Oct 2004 – May 2011
Operating Systems & Integration Engineer
OpenTV · San Francisco, CA & Amsterdam, NL
- Embedded Linux device drivers and networking stacks for global digital television hardware deployed at consumer scale.
- Global software releases and on-site driver integration with vendors and operators across Europe, Australia, and the Americas.
1998 – 2004
Systems Engineer & Hardware Developer
Freelance / SFSU / EngineX / Net.World · San Francisco Bay Area
- Low-level C and ARM Assembly for hardware TCP/IP TOE ASICs (Iready); telescope control and scientific computing at SFSU Observatory; enterprise networking and VoIP infrastructure.
Profile
Brent A. Thorne
Research Engineer bridging physical hardware constraints with high-performance scientific software. Over 20 years across embedded Linux, wireless stacks, computational imaging, and reproducible HPC — with graduate work at UC Berkeley (MSSE) in CUDA, parallel algorithms, and simulation frameworks.
Competencies
- Scientific computing & AI: CUDA, MPI, OpenMP, UPC++, multi-GPU scaling; molecular dynamics (GROMACS, LAMMPS, OpenMM); PyTorch, TensorFlow, SciPy, NumPy
- Imaging, instrumentation & sensing: CT/radio reconstruction, cellular image analysis, automated artifact correction, fluorescence spectroscopy (RS-232/DAQ), V4L2/GStreamer, OpenCV, SDR
- Embedded systems & Linux drivers: C/C++, Embedded Linux, kernel drivers, Device Trees, DMA, RTOS (ThreadX), Yocto, board bring-up, SPI/I2C/UART/CAN, sub-GHz RF, LTE CAT-M
- Mechatronics & field infrastructure: CAD/CAM, CNC, ECAD/PCB, FPGA prototyping (Verilog, Quartus), remote observatories, RF link budgets & telemetry
Education
- M.S., Molecular Science & Software Engineering — UC Berkeley (2026)
- Certificate, Applied Data Science — MIT (2023)
- Non-Matriculated Post-Baccalaureate Studies in Physics & Engineering — UC Berkeley & SFSU (2000–2022)
- B.S., Electronics Engineering Technology — Hamilton Technical College (1993)
Patent
US Patent Application 61834836 — System for Optimal Physical Exercise and Training (Inventor): closed-loop biometric monitoring and adaptive control with real-time bioelectrical signal processing.
Contact