Business Process Automation
Software QA Web Scraping

Overview

Custom-built browser automation and API interception system that extracts detailed influencer metadata (handles, bio links, contact emails, stats) from Influencers Club. Can execute with a UI-based browser panel or via headless backend API parsing.

Key Features

• Automated scrolls and session hooks
• Dual modes: Graphical Web Scraping panel / Console Backend API data
• Clean extraction of verified email lists
• Structured table exports to Excel (XLSX)
• Throttling engine to prevent account flags

Technologies

Overview

Custom background script that intercepts network queries and automates navigation on Collabstr to extract creator price lists, niches, and direct booking details without triggering anti-scraping defenses.

Key Features

• Automated categories navigation & lazy loading parser
• Direct JSON response capture from hidden API endpoints
• Extraction of media price tiers (TikTok, Instagram, YouTube)
• Automated download of portfolio image URLs
• Rate-limit protection via rotating request signatures

Technologies

Overview

A multi-threaded document/data extraction process targeting Modash directories to download creator search directories, handles, emails, and demographic breakdowns in high volumes.

Key Features

• Parallel scraping streams for faster processing
• Extraction of target creator profile data and verified emails
• Dynamic scrolling & search pagination handlers
• Download resume support (avoids duplicate operations)

Technologies

Overview

A highly modular and resilient e-commerce scraper framework built to scrape product descriptions, pricing configurations, seller details, delivery metrics, and reviews across 10+ wholesale and retail platforms (Amazon, eBay, Etsy, AliExpress, Alibaba, Temu, Walmart, etc.).

Key Features

• Sub-domain listing and search crawler
• Multi-marketplace adapters (Amazon, AliExpress, eBay, Etsy, Alibaba, Temu, Walmart, etc.)
• Rotating residential proxy integration to bypass ip bans
• Excel/CSV spreadsheets with structured product properties
• High throughput concurrent download pipelines

Technologies

Overview

One-click Google Apps Script engine that scans any Drive folder, your own or a Shared Drive, and crawls recursively through every sub-folder, and builds a complete, clickable index of every single file inside: PDFs, Sheets, Docs, Slides, images, videos, ZIPs and more.

Key Features

• Recursive sub-folder crawling so nothing gets missed
• Unlimited scale: pagination handles 10 or 10,000+ files
• Dual export to a color-coded Google Sheet and a formatted Google Doc
• Sheet columns: name, type, clickable URL, file ID, parent folder, dates
• Slack-ready Doc: bold names, blue links, dividers per file
• Shared Drive + My Drive support, crash-proof silent error logging

Technologies

Overview

A Python command-line tool that turns LinkedIn prospecting into a guided, semi-automated workflow. It runs a "People" search for a target role and region (Medical Device Engineers around Minneapolis in the demo), opens every matching profile in an automated Chrome session, reads the detected experience level and lets the operator accept or reject each lead with a single keystroke.

Key Features

• Interactive CLI that walks the operator through login, CAPTCHA checks and location-filter setup before the run starts
• Automated "People" search execution and sequential, profile-by-profile browsing
• Experience detection with accept / reject prompts, so only qualified profiles are kept
• Opens each profile in the live browser for instant manual verification when a decision is borderline
• Real-time console log of every profile visited, the decision taken and the connection status

Technologies

Overview

A high-throughput Python scraper that spins up parallel worker streams (W1, W2, W3), each driving its own browser to run Google searches for facility and business names, capture the official website or directory listing, and write the result straight back to the source row. Built for healthcare and senior-living directory enrichment where thousands of rows must be resolved quickly.

Key Features

• Three concurrent worker threads, each with its own browser instance and row range
• Global CAPTCHA stop: the moment Google challenges one worker, all streams pause, the operator solves it once and the run resumes where it left off
• Row-level tracking of the search query, matched URL and destination domain (facility sites, medicare.gov listings, etc.)
• Query-to-row mapping writes results directly into the structured output file
• Live console showing worker IDs, active rows and saved status for every result

Technologies

Overview

A Python automation engine that runs the same category search across hundreds of towns and cities on Google Maps, reading the target list from a plain text file (700+ North Carolina cities in the demo), searching each one for categories such as speech therapy and pediatric care, and saving every verified listing to a master spreadsheet.

Key Features

• City-by-city iteration from simple text lists (nc_cities.txt), so adding a city gets it searched on the next run
• Live Google Maps automation that captures business title, address, phone, website and sponsored status
• Exclusion lists (skip_domains.txt) drop unwanted directories and aggregator sites automatically
• Zero-result cities are logged and skipped, so the run never stalls
• CAPTCHA prompts pause the run for manual solving and continue afterwards
• Spreadsheet-ready export: County, City, Category, Provider, Address, Phone, Website and NPI notes

Technologies

Overview

A post-processing workflow that turns a raw, scraped lead list into a clean master database. A pandas script (work.py) deduplicates the CSV, the cleaned file is uploaded to Google Sheets, and a custom conditional-formatting rule (=COUNTIF(G:G, G2)>1) instantly flags any phone number that still appears more than once.

Key Features

• One-command deduplication from the Windows command prompt, ending with "DONE: leads_deduplicated.csv created successfully"
• Cleaned CSV imported straight into Google Sheets for team-wide visibility
• Custom formula-based conditional formatting colour-codes duplicate phone numbers across the entire column
• Handles 1,700+ rows of County, City, Provider Category, Practice, Address, Email, Phone, Website and Notes in one pass
• Repeatable: re-run the script on every new scrape and the sheet stays clean

Technologies

Overview

A monitoring script (sheet_watcher.py) that attaches to an already-open Chrome tab through the remote-debugging port, watches a live Google Sheet such as a monthly Water Leakage Report, and raises an audible alarm the instant a new row or status change appears, so facility teams react to a leak in seconds instead of at the next manual check.

Key Features

• Attaches to the operator's existing Chrome session, with no separate login or API keys required
• Polls the active sheet tab on a fixed interval and reloads the page automatically
• Detects new rows, edited status cells and missing meter readings against a stored baseline
• Audible + console alarm ("NEW DATA ADDED – ALARM STARTED") that stops automatically once the change is cleared
• Timestamped terminal log of every row count and state change for later review

Technologies

Overview

A lightweight desktop automation script (auto_scroll_ctrl.py) that keeps large lead directories and Google Sheets moving on their own, scrolling on a fixed interval and sending periodic keystrokes so long tables load, render and can be parsed end to end without anyone touching the mouse.

Key Features

• Configurable scroll interval, scroll amount and key-press interval in a three-line settings block
• PyAutoGUI fail-safe: move the mouse to a screen corner and the script stops instantly
• Timestamped log of every scroll and key event for auditability
• Works on any scrolling surface, from web directories to Google Sheets and internal dashboards
• Pairs with the lead tracker sheet (City, Provider Category, Practice, Address, Email, Phone, Website) so thousands of rows can be reviewed hands-free

Technologies

Overview

An automated bot that searches Google Maps for a service category in a target area (pressure washing and exterior cleaning companies in the demo), opens each listing, and harvests the business name, website, physical address, phone number and email into a structured lead file.

Key Features

• Automated search, result scrolling and listing-by-listing navigation on Google Maps
• Extracts name, website, address, phone and email; missing fields are logged as "Not found" instead of breaking the run
• Progress counter ([75/119] Processing) with a saved-entry confirmation for every business
• Handles dynamic map loading and lazy-rendered detail panels
• Runs in a visible or headless browser

Technologies

Overview

A login-protected portal scraper (ascend_scraper.py) that authenticates into the Ascend reseller portal, loads the full publication pricing catalogue, and walks through every listing, saving each unique publication with its niche, pricing and attributes while skipping records that were already captured.

Key Features

• Automated login sequence and session management
• Dynamic loading and parsing of the publication database (1,462+ catalogue entries in the demo)
• Real-time duplicate detection, with "Duplicate skip" or "Saved" logged for every record
• Works with the portal's own search filters, niche tags and paginated tables
• Visible or headless Chrome execution

Technologies

Overview

A watcher script that connects to a running Chrome instance over the remote-debugging port, refreshes the YouTube Subscriptions feed on a loop, stores a baseline of what is already there, and fires a loud, continuous alarm the moment a new upload appears.

Key Features

• Attaches to the operator's logged-in Chrome (--remote-debugging-port=9222), with no API quota and no separate login
• Automated refresh cycle on youtube.com/feed/subscriptions with baseline comparison
• Console status for every cycle: Refreshing page…, Baseline stored, NEW VIDEO DETECTED !!!
• Keyboard controls while running: Q stops the alarm and X exits the watcher
• Infinite alarm until acknowledged, so a new video is never missed

Technologies

Overview

A verification assistant for vehicle inventory audits. The script reads each row of a master Google Sheet (Year, Make, Model, VIN, sticker image URL), opens the door-jamb sticker photo in an automated Chrome window, and asks the operator to confirm the match with a single keystroke, writing the result back to the sheet as it goes.

Key Features

• Row-by-row iteration through the master spreadsheet with the current row shown in the terminal (Row 159, Row 160 …)
• Opens each sticker image URL automatically in Chrome for instant visual inspection
• Interactive prompt to confirm, reject or quit, with "Saved" written back to a status column
• Keeps Year, Make, Model, VIN and Image URL in sync with the sheet
• Resume from any row number, so long audits can be split across sessions

Technologies

Overview

A Python workflow (script.py) that scans an input folder of PDFs such as utility bills, invoices, cheques and accounting notices, reads each file, classifies it into a category such as Utilities, Checks, HAP Adjustment or Ledger Cleanup, renames it to a consistent pattern and writes a tracking row (file name, Slack channel, tagged person, category) to CSV and Google Sheets.

Key Features

• Batch directory scanning with per-file progress ("Processing: scan (18).pdf")
• Text-based classification rules that map document content to business categories
• Standardised renaming (e.g., 19440 ELMWOOD ST_Utilities_2026.01.pdf) for clean archiving
• Tracking output with Slack channel and @person tags, ready for team routing
• Graceful error handling: malformed or unreadable PDFs are logged and skipped, never crash the run

Technologies

Overview

A cleanup script (clean_leads.py) that standardises messy lead exports with repeated header rows, inconsistent phone and fax formats and empty "Preferred Contact" fields, then produces a single clean CSV that imports straight into Google Sheets for collaborative call tracking.

Key Features

• One-command execution from the Windows command prompt with a clear completion message ("DONE: Fax & Preferred Contact fixed. All leads kept.")
• Removes duplicate header blocks and normalises phone / fax formatting
• Keeps every lead, because cleaning never silently drops rows
• Direct import of clean_leads.csv into Google Sheets (separator auto-detected)
• Structured columns: County, Category, Name, Address, Phone, Fax, Email, Spoke To, Notes

Technologies

Overview

An OCR-driven sorting pipeline that reads every scanned PDF in an input_pdfs folder, extracts text with PyMuPDF and Tesseract (fast OCR on the top region of the page), decides where the document belongs and files it automatically. Parsed documents go into date-based output folders (output/2026.03), anything uncertain goes to review/, and every action is logged to output.csv.

Key Features

• Batch processing with per-file terminal progress (Processing: scan (1).pdf, scan (2).pdf …)
• Hybrid extraction: native PDF text first, Tesseract OCR fallback for image-only scans
• Conditional routing into output/<year.month>/ or review/ using shutil
• Standardised file renaming with account number, name, category and period
• Tabular output.csv summary of every processed file for record-keeping

Technologies

Overview

A locally hosted Lead Generation System (v8.1) with a full control panel: create multiple scraping sessions, enter keywords and locations, set speed and safety levels, choose which fields and social links to capture, and watch results arrive in a live table with one-click CSV / Excel export.

Key Features

• Multi-session management, so "electricians in Chicago" and "plumbers in New York" run side by side, each with its own Start, Stop and Reset
• Keyword + location + max-results configuration with "skip first N" for resuming interrupted runs
• Speed presets from Normal to Ultra Fast with a fine delay multiplier, CAPTCHA pause and anti-bot safety controls
• Quality filters (require phone / email, prefer website data, min rating and reviews) and deduplication by name, phone, website or address
• Social link extraction: Facebook, Instagram, LinkedIn, Twitter/X, YouTube, TikTok
• Live analytics (Saved, Skipped, Current Listing, Loaded Listings), session logs and CSV / Excel / Copy export

Technologies

Overview

A scraper for directory tables where key attributes are shown only as icons for image allowed, sponsored, indexed, do-follow, niche flags and price multipliers rather than text. The script walks 1,200+ listings, reads each icon's state from the DOM and converts it into clean Y / blank / "x2.0 cost" values alongside the visible fields (genres, price, DA, DR, turnaround, region).

Key Features

• Icon-state detection (active vs inactive) converted to explicit text flags per column
• Captures genres, price, DA / DR metrics, turnaround time, region and staff / niche indicators
• Side-by-side terminal progress, such as "161/1205 Saved: Ritz Herald News ['', 'Y', 'Y', 'Y', 'Y']"
• Handles pagination and lazy-loaded rows across the full directory
• Writes a structured local dataset ready for filtering and pricing analysis

Technologies

Overview

A locally hosted, multi-session video studio. Paste a YouTube, TikTok or direct link (or drop a file), and VideoStudio ingests the media, lets you cut precise clips by timestamp, transcribes them with Whisper, extracts audio and merges selected clips into a single MP4 or MP3, all from one dashboard.

Key Features

• Multi-session workspaces (Session 1, Session 2 …) so several projects can be prepared in parallel
• Media ingestion from YouTube, Twitter/X, TikTok, direct URLs or local MP4 / MKV / MOV / MP3 / WAV files
• Clip cutting with start / end selectors, a max-duration guard and per-clip preview
• Whisper speech-to-text for a full video or an individual clip; Extract Audio (MP3) in one click
• Merge engine: reorder clips and export as one MP4 video or MP3 audio file

Technologies

Overview

A contact-enrichment tool (email_finder.py) that takes an investor spreadsheet of VC firms, accelerators and angel groups, visits every website, reads the full HTML source plus contact and about pages, and writes back every business email it can find, or a clear status note when it can't.

Key Features

• Batch processing of thousands of target websites from an input CSV / Google Sheet
• Deep email discovery: full page source, contact, privacy and about pages and linked sub-pages, returning all emails found rather than just the first
• Status notes written back per row: "No email found", "Timeout", "Site not reachable"
• Auto-save checkpoints every 10 rows ("Saved till row 40") so an interrupted run resumes exactly where it stopped
• Multi-threaded requests with per-site timeouts to keep long runs moving
• Export to Google Sheets or CSV with the original investor columns preserved

Technologies

Overview

A research and extraction tool for tax-sale and property-auction investors. Running as an overlay panel directly on the USTLA research listings (uls.ustla.com), it pulls every auction listing plus its detail page into a clean, multi-tab Excel workbook organised by month, with county, state, sale date, auction type, venue, opening bids, parcel IDs and property URLs.

Key Features

• One-click list + detail scraping with filters for auction type (Deed, Redeemable Deed, Lien), status and sale-date range, or all pages at once
• Floating in-browser panel with live counters: auctions, details, failed, 429 hits, mismatches and saved
• Retry-failed and resume controls that survive rate limits and page reloads
• Deep field capture: county, state, sale date & time, post date, auction type, sale venue, contact, opening bid, estimated value, parcel / APN and direct USTLA URL
• Excel export (USTLA_Auctions_Filled.xlsx) with one tab per month (Sep 2025, Oct 2025 …), styled headers and auto-fitted columns, ready for pivot tables

Technologies

Overview

A complete operations system rather than a single script, built for an independent construction-consulting firm and documented in a 27-page System Manual. ClickUp is the operational board, Google Drive and Apps Script are the backend (folder tree, workbook, intake forms and release checks), Zapier carries events between platforms, Smartlead and Prospeo run outbound lead generation, HubSpot holds the CRM record and Dropbox is the client-facing file layer. Every engagement moves through one board in one direction, with hard approval gates and an audit log at every step.

Key Features

• Unified ClickUp operational board: every engagement tracked from New Opportunity → Intake → Scope Development → Owner Approval → Proposal → Agreement & Deposit → Field Planning → Field Release
• Google Apps Script backend engine: one project provisions the whole Drive tree (00 Backend, 01 Templates, 02 Forms, 03 Engagements), the master workbook and every intake form, validates responses and runs the fieldReleaseCheck audit
• Zapier multi-bridge layer: one intake submission fans out to four Zaps at once (ClickUp task, Drive folder, Dropbox folder, Engagement Register) and every status change writes a row to the Gate Log
• Automated outbound & lead generation: city-by-city Smartlead campaigns on authenticated, warmed mailboxes with suppression checks and Smart Dialer, plus Prospeo lead sourcing synced into HubSpot
• Structured document lifecycle: per-engagement folders, templated Docs / Sheets, tiered intake modules (Tier 1–3), Site Contact forms and gated proposal / legal release documents
• Fail-safe gate control & audit trail: no stage can be skipped, every backward move is a block, nothing is released without a complete site-contact set and a human sign-off, and the Gate Log's "Evidence confirmed" and "Approved by" columns are never auto-filled
• Operator SOP manual: who does what, daily operation, exceptions, timestamps and do-it-yourself runbooks so the client's own team runs the system without us

Technologies