Pipelines
Raw files in. Trusted tables out. Python pipelines.
$ python run_pipeline.py --source pdfs/

Faith. Tech. Coffee.
PDFs. Notes. Emails. Free text. Screenshots. Data waiting to be extracted, automated, and turned into decisions.
Pipelines, text intelligence, dashboards, and automation — built end to end.
Four layers. One system.
Raw files in. Trusted tables out. Python pipelines.
$ python run_pipeline.py --source pdfs/
Classification and analytics over free text — AI where it adds value, LLMs where they win.
>>> model.predict(texts)
Dashboards designed for decisions, not decoration.
EVALUATE SUMMARIZECOLUMNS( ... )
Power Automate + n8n + Python workflows. Manual work becomes a thing of the past.
flow.run(trigger="daily @ 07:00")
Explore the full portfolio at portfolio.hteller.com
Measure carefully. Extract cleanly. Improve constantly. Precision in the cup. Precision in the data.
Filmmaking sharpened my eye for pacing, framing, and focus — the same discipline behind clean dashboards and useful data products.