Projects

Things that track the work, find the next lead and predict what's coming.

From full apps to quick tools. Here are the ones that matter most to a shop.

100+builds, from small tools to full apps

1,018AI sessions since April 2026

Monthsof work saved by the tools, engines and knowledge bases I've built

Featured

Mally's Home Solutions

The software behind the home-services business I ran over the summer: process management, customer follow-ups, route maps and the data to decide where to grow.

See the write-up

FOCUS

My personal ERP: money, plans and requests in one place, with their status and a forecast of what's next. 263 automated tests.

See the write-up

Research and data

I pick a sector, pull the public data and turn it into analytics and next steps: who to call, where to go and what to do first.

Nov 2026

Trade Door Finder

How I'm finding my apprenticeship. I pulled every licensed electrical, low-voltage and plumbing company in Hartford and Fairfield counties from the state's license data, then mapped and ranked them into a call list: 47 job leads, the hours to each license, and the first three moves to make.

495 companies, 488 of them looked up so far
276 placed at the address the business publishes; the rest at their town's center
Every phone, email and form keeps the page it came from. No guessed emails

PythonCT open dataCensus geocoderMaps

Part of the Mally's app · company details stay private

Hartford County244

Fairfield County251

One square per licensed company. Hartford County: 20 low voltage, 153 electrical, 71 plumbing. Fairfield County: 19 low voltage, 118 electrical, 114 plumbing.

Oct 2026

Closing Watchlist

Finds Connecticut businesses that are closing or for sale and scores each one from 4 to 12, so I can reach the best ones first for equipment, vehicles or routes. 54 leads and 66 listings, rescanned every week.

ResearchLead scoringWeekly scan

Part of the Mally's app · code private

2026

Where to grow

Scores 46 Connecticut towns out of 100 on income, shopping plazas, prospects and competition, then maps them, so expansion is a decision backed by data instead of a guess.

Census dataScoring modelMaps

Part of the Mally's app · code private

Tools and models

Software that turns data into the next action.

Jul – Oct 2026

Lead finder for service work

Finds leads and next actions from data: which towns to work, which streets to knock on pickup day, and who's due a follow-up. Built for blue-collar service work, starting with my own.

PythonJavaScriptMaps

Part of the Mally's app · code private

2026

Travel route optimization

Takes up to 10 routes, rejects bad input, works out each travel time and names the fastest.

Python

Code on GitHub

2026

Membership analysis

Reads a membership file, finds the data problems (blank end dates, confusing codes, dates stored as text) and charts members by status, gender and age.

Pythonpandasmatplotlib

Class project · code private

2026

Small tools

Simple tools that save time. One turns data pasted from Excel or the web into a finished page. One asks eight questions before a build and writes the one-page brief. Plus a schedule builder and an outreach tracker.

HTMLJavaScript

Offline pages · code private

Power-grid AI

The big-picture version of the electrical work I want to learn up close.

Hackathon · 2026

A city's power grid, and an AI that keeps up with it

We simulated Washington DC's power grid: 7 big transmission substations on a 345 kV loop, feeding 51 neighborhood substations. When data centers suddenly pull more power, an AI decides how the grid responds.

58 substations simulated
345 kV transmission loop
1 AI that responds when demand spikes

PythonStreamlitSimulation

Code on GitHub

The amber ring is the 345 kV loop with its 7 substations. Each gray dot is one of the 51 neighborhood substations; the blue squares are data centers.

How I work

I find the slow part of a process, build a tool for it, and check that it actually helped.

Process improvement

Map how the work really gets done, then cut the steps that waste time. At Trumpf that meant fixing data that broke as it moved between Excel, cloud systems and other software.

Tracking tools

Jobs, leads, hours and numbers in one place, each with where it came from, so nothing falls through the cracks.

Predictive models

Forecasts with a likely range, checked against what actually happens. At Leonardo DRS: financial models and price estimates built from past jobs.

I build with AI every day, and I keep a map of every session: 1,018 of them from April to October 2026, by area.

Ventures and business249

School164

Research and reference155

Career and jobs98

Building apps87

How I run my AI tools82

Advocacy role49

Sessions per area. One session can count in more than one area. Personal sessions (268) are left out.

Have room for a part-time apprentice?

I'd like to hear from you. Email or call, whichever's easier.

Starts: January 1, 2027
Hours: 15–20 a week
Schedule: Set together

malachibcollins@gmail.com (203) 383-3089 See my resume