The site says senior-led delivery and no bait-and-switch. That claim is only worth something with a name attached to it, so here it is.
Steve Farmer is the founder of ItsMoreThanSoftware LLC. Every engagement described on this site is scoped, designed, and delivered by senior engineering experience from start to finish. There is no handoff to a junior team after the sale, which is the difference this page exists to make concrete.
The work spans AI agents and orchestration, custom automation, API integrations, data pipelines, and the custom connectors most consultancies subcontract. Those sit together deliberately: agents fail on data far more often than they fail on prompts, so the plumbing and the AI layer are the same job.
Based in Atlanta, GA, working remotely with clients across the US, EU, and APAC.
The site says 30+ years in software and data engineering. Most of that work lives in private systems that cannot be shown here. This is the part that is public.
A self-hosted trading suite for Kalshi weather and inflation markets that I designed, built, and sell. Python on Linux with PostgreSQL state, a shared account-level risk engine, order, fill, and settlement reconciliation against the exchange, and a managed installer with upgrade and rollback. Its public validation page documents release status, known limitations, and what remains unproven instead of a profit scoreboard.
A personal AI orchestration system, still a prototype. Requests become missions, missions are split into bounded tasks routed to agents and models, and sensitive or production changes wait for human approval. The public repository is a sanitized architecture showcase. The implementation stays private.
Posts on DEV from the Predict & Profit build, including a Brier score audit that found one of my own trading models had no forecasting skill, and what replaced it.
Advice that ends at a slide deck is easy to give and worth very little. Every recommendation here maps to something that can be built in your stack, and usually is, inside the same engagement.
A scheduled job, a fixed integration, or a cleaned-up SQL model solves a lot of problems that arrive labelled as AI problems. Saying so costs a project occasionally and is the reason clients come back.
Code in your repository, pipelines in your cloud, models calling your accounts. The goal at the end of an engagement is that you do not need us.
Documentation, runbooks, and training are not a phase that gets cut when the timeline slips. Software nobody else can operate is not finished.
The journal is the closest thing to a portfolio: field notes on Workday SOAP APIs, custom Fivetran connectors, pipeline alerting, and the parts of this work that do not fit in a case study.
Describe the workflow and what breaks today. You will get a straight answer about whether it is worth building.
