I am an engineer in Durango, Colorado. I write software for systems that touch hardware. Test automation, instrument control, embedded and signal processing work, and the data platforms and web applications built on top of them.
Labs I have worked in include Apple, Qualcomm, Samsung, Motorola, Lockheed Martin, and NIST.
The work has covered precision timing and metrology, satellite communications, defense and aerospace embedded systems, cellular and Wi-Fi device bring up, and full stack software and data platforms.
I also run a macro signal platform that ingests, models, and backtests economic data, and I write about signal processing. The macroeconomic dashboard is public, and its charts are interactive.
Available for contract and project work. Two kinds of problem I am a good fit for. A process somebody runs by hand every week, and hardware that needs software written around it.
Six things I get hired to do. Most jobs land on two of them at once. The measurement side frames the problem as a question that can actually be answered. The software side builds the thing that keeps answering it.
Test automation, instrument control, and the applications that sit on top of a bench. Coordinating instruments, running the computation, and presenting the results. I have done this across national labs, chipset vendors, and defense programs, in MATLAB, LabVIEW, C#, JavaScript, and Python.
Embedded and FPGA systems, receivers and transmitters, precision timing, and the DSP underneath. Link budgets and propagation analysis for satellite links. Requirements through validation, including the environmental, EMI, OTA, and signal integrity work that comes with the hardware underpinning the design.
This work usually starts with a recurring manual task. A report rebuilt by hand every week, data copied between systems that do not integrate, or a process that only scales by adding people. I replace those with software that runs on a schedule and talks to the systems directly, and I measure the result in hours per week returned.
ETL pipelines and warehouses on PostgreSQL. Loads re-run safely, and late or missing history can be filled in afterward. The warehouse keeps every version of every record (SCD Type 2, bitemporal), so any report can be re-run as it looked on a past date. Checks at each step, so a number traces back to its source row and a bad one is caught before anyone builds a decision on it.
Next.js and React with TypeScript, Postgres, authentication, payments, and the deployment and monitoring around them. Usually as the only developer, from first requirements through production. I have built and shipped these alone for clients and for employers, and I maintain them afterward.
Python with numpy, scipy, pandas, and statsmodels. Signal modeling, Monte Carlo simulation, and backtesting engines I wrote myself rather than pulling in a framework. On the business side, GAAP income statements, balance sheets, and cash flows with valuation, DCF, NPV, and IRR behind them, for owners raising equity, borrowing, expanding, or selling.
Thirteen macroeconomic and market data sources in one store that records two dates for every number: when it was true, and when I first saw it. A backtest run against it cannot use figures that had not been published yet. Runs continuously on a server I administer.
A live macroeconomic dashboard tracking growth, inflation, labor, rates, liquidity, and volatility, with every number traceable back to the release it came from.
A Squarespace site rebuilt as a Next.js product: accounts, Stripe billing, a moderated community forum, and a gated video library. Content is MDX validated against JSON Schema at build time, and the legacy URLs were reconciled against redirects so the site kept its search ranking through the move.
More, including the writing, on the projects page. See all projects
Describe the problem and what you have tried so far. I read everything and reply.