GTM / Lead-Gen Engineer

Pipeline, built from intent.

I'm Ayush Kumar. I build the outbound systems behind other people's campaigns: clean sourcing, cheap enrichment, and buying intent turned into booked replies.

Book a call Connect on LinkedIn

Available for a full-time GTM role · open to contract projects

0 leads reached
0 positive replies
0 months building GTM systems
Claude Code not just Clay

Profile

A seller who went and built the system.

Intro · who I am and how I work

Open to work Looking for a full-time GTM engineering role where I own pipeline as a product. Open to contract projects too.

Ayush was a pleasure to work with. He provided a thorough perspective as he scoped the project, asked plenty of questions to understand the best possible outcomes (& data usage), and delivered the project in a timely manner. He was a clear and frequent communicator, and made it easy to move forward together. Would definitely recommend working with Ayush.
Joe Hinderstein Joe Hinderstein Client

Approach

Intent-led, not spray-and-pray.

Volume is easy; relevance is the work. I start from who's actually in-market, keep the list clean, and let signal decide who gets a message and what it says.

  1. Source on signal

    Find the right accounts with Apollo, Ocean.io and custom Claude Code scrapers, then dedupe against everyone we've already touched.

  2. Enrich clean & cheap

    A custom email waterfall and a Supabase cache mean we never pay twice for the same person. Quality up, cost down.

  3. Write tight

    Two-line emails, personalized from each prospect's ICP and business, variant tested against variant. Then double down on what replies.

  4. Reply fast

    A reply agent catches positive responses and keeps conversations moving, deliverability watched throughout.

Results

A campaign, managed end to end.

I ran outbound end to end for a lead-gen agency: sourcing, enrichment, sequencing, testing, replies.

0 leads reached
0 positive replies
0 sequences + follow-up
2-line emails, nothing longer
  • Sourced with Ocean.io and custom logic in Claude Code.
  • Personalized every prospect with Firecrawl + DeepInfra: their ICP and what the business does.
  • Five short sequences plus follow-up, copy kept to strict two-liners.
  • A/B tested, doubled down on the winner, paused the rest.
  • Positive replies handled by a reply agent, deliverability watched throughout.
Walkthrough: the campaign results, start to finish.

Build · Clay bootcamp

A cluster scraper and a cost-aware enrichment engine.

Built with Claude Code, not by living inside Clay. A whole lead list from a custom scraper, and a Clay table that treats enrichment cost like a budget to defend.

01SourceApollo data in; dedupe against past campaigns.
02RepairNormalize company; find missing websites and merge them back.
03Cache checkSupabase lookup skips anyone already enriched for another client.
04Email waterfallApollo → Blitz → LeadMagic, each verified by Reoon + Bounceban.
05PersonalizeGenerated PS line, sub-industry and three tailored ideas per lead.
→Clean listSaved back to Supabase so the next client never pays twice.
Walkthrough: the scraper and the Clay enrichment table.

Selected work

Builds, walked through.

Short screen-recorded walkthroughs of systems I've built recently.

Scraper · Claude Code + trigger.dev

Buyer intent scraper

Timing decides outbound. A company that raised yesterday is deciding now; one that raised months ago already chose its tools. So this runs every morning, sweeping 15+ public sources (Product Hunt, Crunchbase, Reddit, G2, TechCrunch and more) for live signals: funding, outbound hiring, launches, competitor switches, deliverability pain. Every hit is enriched, ICP-scored 0 to 10, and given a 50 to 200 word “why now” before only qualified B2B leads reach the CRM.

  • 700+ companies scraped daily, 50 to 100 qualified leads per run
  • 2-3% reply rate on v1 with zero manual effort (targeting 8-10%)
Agent · low-cost AI infra

LinkedIn copywriting agent

Staying active on LinkedIn eats hours a week: writing posts, researching top creators, finding ideas, keeping a schedule, checking analytics. This folds the whole workflow into one platform. It studies top creators and extracts their patterns, generates posts in your own voice, runs topic and keyword research, supports media, plans content, and tracks post performance, all on low-cost AI infra (NVIDIA, OpenRouter, Together AI, Exa, Tavily, Supabase).

  • Hours of research and writing cut to minutes
  • Consistent posting at near-zero daily effort and cost

Stack

The stack I build with.

Build & automation

  • Claude Codeagents, scrapers, pipelines
  • Clayenrichment tables & routing
  • n8nself-hosted workflows
  • Makescenario automation

Data & sourcing

  • Apollocontact & company data
  • Ocean.iolookalike sourcing
  • TrigifyLinkedIn signal scraping
  • Firecrawlweb scraping for enrichment

Enrichment & email

  • Prospeoemail finding
  • Blitzemail finding
  • LeadMagicemail finding
  • DeepInfraLLM personalization
  • Supabasededup & cost cache

Deliver & CRM

  • SendKitsending infra & deliverability
  • Reoonemail verification
  • Bouncebanemail verification
  • AttioCRM

Contact

Hiring a GTM engineer? Let's talk.

I'm looking for a full-time GTM engineering role, and I'm open to contract projects. Building pipeline as a system, or just want to compare notes? Reach out.

Book a 30-min call ayushkr7894@gmail.com LinkedIn