I build custom operational projects—and engineer agentic improvements over them.
I am an engineer with accounting training. I build systems for owner-operators who are tired of manual work and cannot afford data that doesn't reconcile.
The Paper Registers of a 27-Loom Weaving Mill
My introduction to operational software didn't happen in an abstract tech incubator—it happened standing on the loud, humid weaving floor of a commercial textile mill.
The mill had 27 industrial looms running day and night. Everything was tracked on physical, grease-stained paper registers. Shift supervisors wrote meterage numbers with ballpoint pens. Clerks transferred those numbers into multi-user Excel files on shared hard drives. Every Friday evening, two accountants sat down with calculators to reconcile yarn consumed against greige fabric produced.
The numbers never matched. Someone had inevitably overwritten a cell formula, transposed a lot number, or miscalculated piece-rate wages. The cost wasn't theoretical: it was hundreds of thousands of dollars in untracked yarn drift and days of lost clerical time.
I built TextileMode ERP to solve that exact operational problem. We didn't deploy an off-the-shelf chatbot or generic SaaS. We built a custom relational ERP from scratch that modeled their physical looms and double-entry accounting. Then, we layered agentic intelligence over it—automated shift briefings, yarn shortfall alerts, and variance detection. It has now been running continuously in production with real financial data for over 10 months.
How I Think • Engineering Principles
Build the custom system first. Layer agents on top.
An AI agent without a solid database is just a hallucination waiting to corrupt your books. We build custom operational systems from scratch—relational schemas, strict constraints, and role-based permissions—and engineer agentic workflows over that robust foundation.
Data must reconcile to the cent.
If an operational system cannot produce a trial balance where debits equal credits to the exact penny, it is not operational software. I design every schema with strict relational integrity constraints and double-entry accounting mechanics.
Floor reality beats abstract software theory.
Software must work for the supervisor in work gloves on a dusty factory floor, not just an executive on a pristine laptop. If the interface is awkward to use during a shift change, workers will revert to paper logs.
Zero vendor lock-in. Full client ownership.
You own your git repository, your database encryption keys, and your cloud hosting accounts. I work via scoped IAM developer accounts so you can inspect, audit, or transfer the code anytime.
I reply within 24 hours on business days. No account managers or ticketing layers; you communicate directly with the engineer building your system.
Every Friday by 17:00, you receive a concise written report with a working staging demo link, completed milestone deliverables, and any blockers for the upcoming week.
Verified Credentials & Training
Truth Gate VerifiedRigorous foundations in C, Python, SQL, algorithms, and memory management.
Production architecture in Forward Deployed Engineering and deterministic agent pipelines.
Mathematical balance assertions, transactional journals, and auditable reconciliations.
Want to discuss your operations?
Book a 20-minute discovery call to discuss the custom system you need.