The latest articles on Forem by Dave Patrick Bulaso (@patrickpatrick27). https://forem.com/patrickpatrick27 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3848585%2Fba0c29c8-f07d-45f5-b9e4-33a78773c3c3.png https://forem.com/patrickpatrick27 en Dave Patrick Bulaso Sun, 29 Mar 2026 03:22:18 +0000 https://forem.com/patrickpatrick27/how-i-built-a-247-lead-generation-pipeline-with-python-and-claude-code-at-21-18mm https://forem.com/patrickpatrick27/how-i-built-a-247-lead-generation-pipeline-with-python-and-claude-code-at-21-18mm <p>I'm 21, based in the Philippines, and I work as a software engineer for a UK consultancy. One of my tasks was building a lead generation system that could find and enrich business profiles automatically.</p> <p>Instead of doing it manually, I built a pipeline that runs 24/7 on a $150 mini PC sitting in my room. Here's how.</p> <h2> The Problem </h2> <p>My client needed enriched profiles of C-suite executives — names, companies, roles, LinkedIn URLs. Doing this manually would take forever.</p> <h2> The Stack </h2> <ul> <li> <strong>Python</strong> — all scripts</li> <li> <strong>Playwright</strong> — browser automation for scraping</li> <li> <strong>Claude Code</strong> — my primary development tool (I built the entire system with it)</li> <li> <strong>Beelink Mini-S</strong> — a tiny Ubuntu server running 24/7</li> <li> <strong>Docker</strong> — for containerized services</li> <li> <strong>Google Sheets API</strong> — for syncing data to the client's spreadsheet</li> </ul> <h2> How It Works </h2> <p>The pipeline runs in stages, each as a separate script:</p> <ol> <li> <strong>Scrape</strong> — Pull company and executive data from public sources (Companies House API, corporate websites)</li> <li> <strong>Clean</strong> — Deduplicate, normalize names, validate data</li> <li> <strong>Enrich</strong> — Cross-reference with LinkedIn and other data sources</li> <li> <strong>Sync</strong> — Push to Google Sheets via API</li> </ol> <p>Each stage runs independently. If one fails, the others keep going. The whole thing runs via cron jobs on the Beelink.</p> <h2> Results </h2> <ul> <li>Built a pipeline that generates <strong>hundreds of profiles per day</strong> automatically</li> <li>Runs <strong>unattended 24/7</strong> while I sleep</li> <li>Total infrastructure cost: ~$3/month in electricity</li> </ul> <h2> What I Learned </h2> <p><strong>Claude Code changed how I work.</strong> I didn't write this system line by line. I described what I needed, Claude Code generated it, I reviewed and refined. My output speed is probably 3-5x what it would be without it.</p> <p><strong>A $150 mini PC is a legitimate server.</strong> It runs Ubuntu, Docker, cron jobs, Python scripts — everything a cloud VPS does, but with no monthly bill.</p> <p><strong>Automation compounds.</strong> The time I spent building this system paid for itself within the first day. Everything after that is free output.</p> <h2> My Setup </h2> <p>I'm currently running:</p> <ul> <li>Lead generation pipeline (Python/Playwright)</li> <li>Job monitoring bot (scans 3 platforms every 5 minutes)</li> <li>Google Workspace integrations (Calendar, Gmail, Sheets)</li> <li>AI assistant infrastructure (OpenClaw + Claude Code)</li> </ul> <p>All on one mini PC, all 24/7, all from my room in the Philippines.</p> <p>I'm Dave. I build systems that run without me. If you have questions about automation, Claude Code, or running infrastructure on cheap hardware, ask in the comments.</p> <p><a href="https://github.com/patrickpatrick27" rel="noopener noreferrer">GitHub</a> | <a href="https://patrickpatrick27.github.io/Flutter-Web-Portfolio/" rel="noopener noreferrer">Portfolio</a> | <a href="https://linkedin.com/in/dave-patrick-bulaso-169b7b307/" rel="noopener noreferrer">LinkedIn</a></p> python automation openclaw claudecode