🛠️ INTERNAL TOOLS BUILT + AUTOMATION

Tools that save hours and replace busywork. Built fast and used in production.

Automated Photo Archive & Search Portal

25 Years of Images...

A zero-server pipeline that crawls job folders weekly, classifies photos with a CNN into Full Tank / In-Process / Details, generates previews, and publishes a static, searchable portal on the company’s network drive.

Across ~ 1,400 folders (2000-present)

Scope: 38,296 photos archived to date from ~1400 distinct job folders; originals remain untouched in source directories; updates are idempotent and scheduled via Windows Task Scheduler.

SQLite Searchable Data Base

Full-text search (FTS5) over file names + extracted text with AND/OR and NEAR queries; year/job/customer filters and file-type filters.

A desktop search engine that indexes 210k+ files / 2.4k jobs and returns job-level results instantly.

Supports full-text (FTS5, AND/OR/NEAR), smart ranking (bm25 top-K), and filters for file-types.

UI is chunk-rendered to avoid freezes;

DB is opened read-only & immutable for zero-risk reliability.

Stack: Python (Tkinter), SQLite FTS5

CUSTOM ESTIMATOR'S APP CUSTOM WELDER RESUME SCORING APP

💬 “Hell yeah! Now my assistant can do estimates as quick as me. Can you build me a tool for cone bottoms next?” - Sales Manager

🎯What it solves:

Estimators were manually pulling specs from 70+ page semi-structured PDFs.

🛠️ What I built:

Streamlit app that extracts key specs from API-650 tank calcs.

⚡ Cuts quoting time by ~33% and reduces manual errors.

🔧 Tech: Python, pdfplumber, pandas, Streamlit

🟡 GitHub Repo | 🟢 Live App

💬 “I hate digging through résumés. This thing is badass. I can see who's worth calling in seconds.” -Hiring Manager

🎯 What it solves: Hiring manager needed to screen 60+ welding résumés fast — current process was slow and manual.

🛠️ What I built: A PDF résumé reviewer with custom scoring logic (welding type, materials, blueprint/math knowledge).

⚡ Built in 1.5 days.

🚩 Flags test-readiness, exports Excel summaries.

🔧 Tech: Python, Streamlit, PyPDF2, pandas, regex, dateutil

🟡GitHub Repo | 🟢Live App

CONCENTRIC CONE MATERIAL & LAYOUT ESTIMATOR

🎯What it solves:

Estimators were using AutoCAD to calculate material for cone bottoms — a slow, overkill process for quoting jobs.

🛠️What I built:

App that calculates optimal plate layouts for truncated cones using:

sector geometry
kerf + nesting logic
real-world plate constraints

Outputs break diameters, waste, and 2D layout — no CAD required.

Result:

⌛ Saves ~3 hours/week 📉 reduces plate waste. Currently in testing at Savannah Tank.

Tech used:

Python, Streamlit, sector math, pandas, matplotlib, domain logic, optimization logic (nesting/rotation)

🟡 GitHub

🟢 Live App

Page updated

Google Sites

Report abuse