AI & automation engineer · Houston, TX
Brandon Howell
I built an auto finance group's operations and AI platform on my own, from an empty repo to production in seven months.1
- scheduled jobs
- 130+
- tools in one AI assistant
- 46
- automated report deliveries
- 45+
- vendor integrations
- 12
Sample data · tap a question3
Selected work
Each piece runs in production for a multi-location auto retail and finance group.
Operations platform
One platform for sales, collections, inventory, marketing and finance.
Built alone in seven months for a $75M+, 4,500-account portfolio. 130+ scheduled jobs, 45+ automated report deliveries and a dozen integrations.
Explore the platform →Job runs by hour, a typical week. Illustrative.2
AI analytics assistant
Ask the business a question in plain language. 46 tools answer from live company data, with the cost of every answer tracked.
Try the assistant →Daily operations and financial reports
Every morning, each key number is graded against its own last 30 workdays, with an AI-written summary on top.
Open a report →Skip tracing workbench
One click searches six sources for a borrower who can't be reached, and Claude writes the dossier and next step in about 5 minutes. By hand, it took an estimated 4–6 hours.
Run a search →Grounded AI summaries
The model words numbers computed in advance. A validator rejects any account it wasn't given.
Try the validator →AI at cents per task
Prompt caching and batch processing. AI research reports at about 20 cents each.
See the costs →Data you can trust
Found a collections report that overstated work six-fold, and rebuilt it on real contact records.
See the checks →Predictive models
A promise-to-pay model from 227,789 past promises, within about 6 points in backtesting.
See the models →A dozen integrations
Including GPS telematics for 8.5k devices on 5 nightly pipelines.
Tech specs →Security built in
Role-based permissions, audit logs of logins, exports and data reveals, and encryption of identity data.
Tech specs →Earlier work
2022 – 2026
Futures trading analytics
Python ETL over 50M+ rows of futures market data, a PostgreSQL time-series store, a backtesting system, and a performance framework tracking 12+ metrics. Funded at three proprietary trading firms.
2019 – 2022
Hospital process engineering
Texas Children's Hospital. Lean Six Sigma projects, the model behind the US News rankings work across 10 specialties, and a staffing forecast for COVID-19 testing.