// SOFTWARE AND DATA INFRASTRUCTURE FOR SCIENCE
I build the software and data infrastructure underneath scientific and engineering work, so failures get caught the same hour and results can be rerun by someone who wasn't in the room.
// THE PROBLEM
The expensive problems in research software are rarely the crashes. They are the jobs that log a success and produce nothing, the numbers nobody can reconstruct six months later, and the system that only runs because one person remembers how it works.
That is the work I'm hired for: finding where a system fails quietly, then building the monitoring, automation and provenance that keep it honest.
Jobs report success while the output is missing or wrong.
No one can rerun the analysis that produced the figure.
The pipeline lives in one engineer's head.
// SERVICES
Every engagement is quoted against a written scope before any work starts. Ranges below are typical, not final.
Data management, software sustainability and computational approach sections written to the agency solicitation, plus one revision round. Also available as a technical review of a draft.
Find where the system fails silently, then fix it: health checks, alerting, stall detection, auto-restart, single points of failure.
Provenance capture, versioning, environment pinning, documented formats, and a working path for an outside party to rerun your results.
Build and test automation, regression coverage on numerical output, dependency hygiene, and getting code out of one person's head and into a repository.
Phishing and voice phishing simulation, employee risk scoring and awareness program design.
Covers the human attack surface and the engineering side of technical controls. Does not cover CMMC or NIST SP 800-171 compliance, System Security Plans, POA&Ms or assessment readiness.
// SELECTED RESULTS
120:1
Cut in worst-case failure detection time, from up to five days undetected to same-hour alerts
53%
Lower forecast error (RMSE 134 to 63 km/s) from a real-time correction layer on a physics model
10+
Reliability and maintainability defects resolved across a six-phase remediation
2.5 yrs
Building a full-stack AI voice phishing simulation platform used by multiple client companies
Case 01
Space weather forecasting, NASA-funded. Led the compute and reliability overhaul of a space weather forecasting pipeline running on NASA's Pleiades supercomputer, in support of Artemis. Closed the monitoring gaps that had let production failures run undetected for days, fixed an unmonitored single point of failure and a job-dispatch race condition, and brought production scripts under version control.
Case 02
Enterprise security platform. Built a Python platform running AI voice phishing simulations, with an ML pipeline scoring employee vulnerability and clustering behavioral response patterns.
// FOR SBIR TEAMS
Most proposal consultants are strong on narrative and commercialization. I'm the technical side: the data management plan, the software sustainability section, and the computational approach, written by an engineer who builds these systems and reads them the way a technical reviewer will.
I also work with proposal firms on a subcontract or white-label basis, and with state programs, SBDCs and I-Corps hubs that refer companies to outside technical help.
Some SBIR and STTR awardees can use Technical and Business Assistance (TABA) funds for outside technical help. Eligibility and amounts depend on the agency and the award, so check your solicitation.
// PROCESS
20 to 30 minutes to understand the system or proposal and what "done" looks like.
A clear deliverable, a price and a delivery date, agreed before any work starts.
The work, documented so your team can own it after I'm gone.
For a first engagement, I'm glad to start with one narrow deliverable at the low end of the range, so the work speaks before anything larger is committed.
// ABOUT
I'm an engineer who works on NASA-funded research, currently on space weather forecasting infrastructure running on the Pleiades supercomputer. What I'm hired for is the software and data infrastructure underneath the science: pipelines, monitoring, automation, and the reproducibility layer that lets someone else trust the results.
Before that I spent about two and a half years building an AI-driven security awareness platform for enterprise clients.
Based in Peoria, Arizona. Working remotely with teams nationally.
// CONTACT
Email is fastest. I reply within one business day.
Peoria, Arizona · Remote, available nationally