Off-highway controls
Diagnostics · Internal tools
The engineer who already lives the diagnostic loop — then built the tools so it would not live in inboxes.
For seven years at Cummins I have sat in the gap between a field symptom and a defensible engineering answer. A complaint arrives. Evidence is incomplete. Calibration, hardware, and software all look plausible. The wrong next step wastes a technician’s day and a customer’s machine.
I lead that loop at scale: visibility across more than 600 team-level off-highway field issues since 2020, 107 tickets assigned to me in the current system — every one closed complete — and Datalinks SME ownership across programs. I designed the intake, triage, Level 1–3 RACI escalation, and source-backed discipline the work actually requires.
I built three internal applications that attack three different layers of the problem: deterministic J1939 parsing, workflow orchestration, and searchable, source-traceable lookup. Trace Converter taught me that parsing beats folklore. Zenith taught me that service knowledge lives in state and routing, not inboxes. OFC Assist taught me that an answer an engineer cannot trace to a source is an answer they will not trust.
B.S. Computer Engineering, FAMU-FSU College of Engineering, Florida State University, 2019. Promoted through four roles of increasing scope from rotational Controls Engineer to Controls Lead for B4.5 and X15.
01 · In regular team use
A local-first diagnostics knowledge product. Fragmented J1939, fault, Calterm, service, and emissions sources become one searchable surface where every hit is traceable to its authoritative page. An answer without provenance is not an answer.
02 · Alpha through beta-ready
Email and spreadsheets were the system of record. Zenith made program-and-task state, ownership, deadlines, notifications, dashboards, and migration into one version of the truth — with a closed user-feedback loop (30+ items closed, 11 formal submissions).
03 · First tool
Expert-only CAN-trace analysis, automated: PCAN traces and J1939 symbol files in, bit-level SPN extraction across byte boundaries, PDU1 normalization, TPCM handling, duplicate suppression, deterministic SPN-support classification out. Metadata-driven, so new controller configurations do not require a rewrite.
Symptom, urgency, machine impact, available evidence, ownership. Same-day triage as the default, not the exception.
Fault codes, calibration, datalogger / datalinks, tests, field history. Separate the complaint from the cause.
Service documents, calibration records, prior tickets, engineering notes, specialist knowledge. If it cannot be cited, it cannot be trusted.
Current vs prior calibration. New vs newly referenced. Hardware vs software. Bench when the field cannot decide.
Level 1 known. Level 2 incomplete or program-specific. Level 3 novel, safety, cyber, architecture — human authority.
Write the rationale, not just the answer. Feed the pattern back into tools, forums, and the next engineer.
Naples, FL · Remote-ready · Resume PDF on this page. This site is a personal advertisement, not a Cummins property.
I have spent seven years turning messy field symptoms into owned next steps, and three tools into regular team use. If that is the kind of operator you need on a controls, diagnostics, or service-engineering problem, let’s talk.
