Background
For two decades I have operated an ecommerce business and built much of the software behind it. That work ranges from a custom Django storefront that has remained in production for about 15 years to newer systems for marketplace accounting, dynamic pricing, purchasing, advertising optimization, shipping, returns, customer service, and operational monitoring.
Running the business changed how I approach engineering. I have been the person waiting for a marketplace settlement, deciding which inventory to buy, investigating a reconciliation difference, paying for advertising, and needing a carrier label to print now. I know that a technically elegant system is not useful if an operator cannot understand its decision, recover from a failure, or trust its numbers.
I am comfortable with the unglamorous details that make systems durable: conflicting identifiers, late-arriving events, idempotent external actions, pagination, stale credentials, barcode normalization, FIFO layers, partial settlements, printer formats, cron failures, and migrations that have to coexist with old production code. I prefer clear constraints, durable state, audit trails, reconciliation, and small reversible steps.
Applied AI
I also build applied AI systems. AgentLib, my lightweight Python agent framework, and Code Agent, my interactive coding agent built around a Python REPL, are the foundation of much of my newer work. I use them for pricing, classification, research, self-healing browser automation, and development. My production approach keeps identity, money, state transitions, and side effects deterministic while agents handle bounded judgment. I evaluate models and prompts against representative cases and consider cost as well as quality.
Beyond ecommerce
I built a scheduling/check-in and client-outreach platform for a law firm. That work reinforced the same lesson: useful software starts with the operation. Receptionists need fast check-in and accurate wait-state transitions; outreach staff need documented client facts rather than speculative personalization; administrators need append-only history and recoverable records.
What I am looking for
I am exploring my next chapter. I am primarily interested in a steady engineering role where I can understand the business deeply, build software that matters to daily operations, and keep improving it over time. I would also consider an early-stage team where practical judgment and broad technical experience are needed. Focused contract work may be a fit as well.
Likely fits:
- Senior backend or systems engineer for internal tools, operations platforms, or integrations
- Early engineer working closely with operators and founders
- Practical engineer building applied AI into everyday workflows
- Leading a small engineering team
- Focused contract or advisory work
Where I do my best work
- systems that real users depend on every day
- messy integrations and real migrations
- working directly with the business and operations
- small, focused teams
- AI tested against real cases
- reliable software built to last
Contact
The best next step is a conversation about the operation you run and the systems it depends on.