Systems I've built

A working sample of the operational software built for one customer-success team. Every impact figure here is traceable to a measurement; where there isn't a verified number yet, none is shown.

Customer-success command center

A live dashboard for tracking customer health, at-risk flags, renewals, and team workload in one place, replacing scattered spreadsheets.

33 at-risk customers tracked to an outcome (21 retained, 4 churned, 8 open)

Python, a lightweight web dashboard, spreadsheet-backed data model

Read the full case study →

Market intelligence pipeline

An automated research pipeline that collects public market data on a schedule, scores it for growth signal, and hands back a ranked opportunity report, no manual pull required.

4-stage pipeline: collection, scoring, synthesis, ranked output

Python, public API scraping, SQLite, spreadsheet export

Read the full case study →

Email drafting with voice calibration

Drafts outbound customer emails automatically, then learns from a person's edits so the drafts converge on their actual writing voice over time.

Measured 45.9% run-time reduction on the drafting pipeline (shadow corpus, live confirm pending)

Python, a calibration loop, prompt-based drafting

Value / ROI engine

An internal accounting layer that credits the customer-success system with the revenue it actually retains, so the team's impact is a number, not a narrative.

$30,590 in retained revenue credited to the workflow (21 retained customers)

Python, a spreadsheet-backed ledger, event sourcing

Price history tracker

Built for a small coffee roaster whose supplier sends a new price sheet every couple of weeks, each one a standalone spreadsheet with no memory of the last one. Stitches every sheet into one running, item-level price history that updates itself.

5+ items tracked across every price run, refreshed with one click

Spreadsheet-native data pipeline, folder-based ingestion, auto date-stamping

See the project →

Data-entry / form-fill automation

Automates a repetitive multi-field data-entry task that previously required manual copy-paste across systems.

Measured 73% time reduction on the form-fill step (shadow corpus, live confirm pending)

Python, browser automation, a dry-run verification harness

Video-to-documentation pipeline

Takes one monthly walkthrough recording and turns it into a full set of polished, branded release documents automatically.

4 deliverables generated per release: detailed guide, interactive guide, a reassign-strategies guide, and a poster

Python, transcript extraction, templated document generation

Discord support bot

A chat-based support bot backed by a real knowledge base, so common questions get answered instantly and correctly instead of waiting on a person.

1 bot, built across a 6-stage pipeline, backed by a maintained knowledge base

Python, a knowledge-base index, Discord integration

Support-hub SOP library

A structured library of standard operating procedures and quick-start reference cards, one per support channel, so new team members and customers get consistent answers.

18+ channel quick-start cards plus a full SOP set

Markdown-based docs pipeline, a validation gate before publishing

Personal knowledge wiki

A searchable internal wiki that organizes reference material into a browsable, linked knowledge base instead of scattered notes and documents.

Static site generation, linked markdown pages

Screen recorder app

A lightweight native screen recorder, built as a simpler, faster alternative to general-purpose recording tools for quick walkthroughs.

Native desktop app

Skill / automation framework

A framework for turning a manual process, once it's been done a few times by hand, into a repeatable, validated automated skill.

Modular task specs, a validation gate for each new automated skill