Growth Engineer
I worked on investment banking go-to-market, sell-side and buy-side mandates. My job was to make the pipeline readable and get outreach performing at volume.
The problem
The deal teams worked out of spreadsheets and memory. Lists went stale, personalization was all manual, and nobody could say which campaign produced which conversation.
What I built
The internal tooling: AI-assisted tools and React dashboards on top of our CRM data. Pipeline tracking, campaign orchestration and the analytics the team opened every day.
- Lead-list scoring and validation pipelines, with a written reason behind every score.
- Email and LinkedIn sequences run side by side, with personalization that stayed cheap at volume.
- Signal pipelines for M&A and competitor acquisitions, built off government filings, plus Crunchbase scrapers for the companies those miss.
- A personalization agent that wrote from what we knew about an account rather than from mail-merge fields.
How it worked
Everything started with the source - government filings, Crunchbase, Apollo, LinkedIn - then went through Clay for waterfall enrichment, dedupe and validation before it was allowed near the CRM. Only records that cleared the scoring bar went into a sequence, and sequences branched on signal type, so a company that had just announced a deal got a different opener than one that had gone quiet.
Replies got routed as soon as they landed, and every step wrote back to the CRM. That kept the dashboards accurate instead of turning them into a second, prettier version of the truth.
What came out
- Flagship campaign: 2,376 sends, 95 replies (8.42%), 52 of them positive - bounces under 1%.
- Volume campaigns: 4,000+ sends each, still landing 6-10% reply rates.
- LinkedIn: 356 HeyReach requests, 51% accepted, 62% of messages answered.
- PlusVibe: 18,000+ leads contacted at a 1.7% bounce rate.
- The dashboards ended up in the team's daily routine, and the scoring model shipped with documentation so it kept working after I left.