GTM systems engineer · currently at Otera

Outbound,
engineered.

I build the machine that books the meetings: domains, data, signals, and the agents that run them. Then I hand it to your team and it keeps running.

See the campaigns

CURRENTLYGTM Engineer, Otera BASED INMumbai, working globally OPEN TOGTM engineering roles

Vaibhav Aher — EXHIBIT ABuilds the machine.
Hands over the keys.

01 / Who I am

Solving revenue problems,
one agentic loop at a time.

I'm a GTM engineer. I believe revenue problems are systems problems.
So I build agentic systems, one loop at a time.

A loop is not a dashboard or a deck. Something watches for a signal, qualifies the account, writes the message, sends it, reads the reply, and feeds what it learned back to the start. I build that from zero, domains, data, agents and the code underneath, then hand it to the team and it keeps running. Every number on this page is a live pull from the platform that sent the email, or a screenshot of it.

Start from the revenue problem.Build the loop, not the deck.Kill on the data.

02 / Where I've shipped

Production code,
tied to revenue.

Three roles, four fractional clients.
Read left to right, newest first.

05/2026 — PRESENT
Otera

GTM Engineer · Remote, Innsbruck · Series A agentic automation (Allianz, Siemens)

Built a production GTM app in Next.js, TypeScript and Supabase/Postgres, used daily by a 20+ person GTM team: a full ABM monitoring stack where AEs get live company- and person-level signals, and agents with account context draft the messaging.

Designed the in-house signal framework that replaced an off-the-shelf ABM stack: n8n and Python workers ingest RSS, podcasts, YouTube and LinkedIn conversations, LLM-classify them into intent clusters, and push qualified accounts into Clay. Claim-batch RPCs and run tables make every run idempotent and observable.

Built the CRM underneath it, the entire TAM scored against the ICP by a versioned LLM qualification pipeline on a 30+ table Postgres model, plus an MCP server so the team's AI tools can read it.
OutputOne platform, 20+ daily users, shipped to production.
08/2025 — 06/2026
BetterFutureLabs

GTM Engineer · Remote, Hoboken · venture studio building agentic AI products

Set up cold email infrastructure from scratch (Instantly, EmailBison, Heyreach) and wired the whole GTM stack together with Claude Code, n8n and Apify.

Shipped a brand OS for AI agents: a structured layer of positioning, voice and visual rules that content and outreach agents build on, so every output stays on-brand.

Automations watching job posts, new hires and Reddit, Facebook and X threads for live buying signals; Clay qualifies and enriches each account before outreach.
OutputBuild with Wave: $0 → $25k+ MRR on outbound and founder-led content.
03/2025 — 08/2025
Ateliex

GTM Automation Operator · Remote, Sydney · SaaS affiliate marketing agency

Grew affiliate revenue for Rise, Sudowrite and BetterPic: the full acquisition path from lead sourcing to conversion tracking, feeding a B2C pipeline.

n8n automation connecting 15+ platforms (HubSpot, Clay, Attio, Customer.io) over REST, replacing manual affiliate ops with workflows that ran themselves.

RAG agents on Supabase and Pinecone for email automation and affiliate support, and a 24/7 intent capture that fired sequences off real-time prospect activity.
Output$100k+ MRR affiliate pipeline, run on n8n.
ALONGSIDE — FRACTIONAL
Four clients

Fractional GTM Engineer · outbound built from zero, handed over running

Skydive (Series A). Outbound motion from scratch. Message-market fit proven across 20 campaign cells, 15,029 emails, 398 replies.

Pemo (Series A, UAE). Built the UAE motion from nothing and scaled it to 25+ meetings a month.

Houses of Growth. The AEO team behind Warmly and Jamie. I worked the distribution side and built their list intelligence.

Revenue Flow. GTM partner. Built the stack that automates the agency's own fulfilment and demand chain.
EducationB.Tech, Communication & Computer Engineering, LNMIIT, 2021–2025.

03 / The campaigns

Real plays.
Real numbers.

Performance and method only.
The copy stays with the clients who paid for it.

● ● ●all campaigns / five clients / 2026replies, not opens

#	Client	Who, what	Cells	Sent	Leads	Replies	Reply rate	Positive	When	Source
01	Skydive skydive.com ↗	B2B startups, 15–50 people · AI teammates	20	15,029	8,204	398	4.85%	35 interested, floor	Jun – Sep 2026	Smartlead API
02	Pemo pemo.io ↗	UAE SMBs · corporate cards	6	11,618	6,125	122	1.99%	53 interested	Jul – Sep 2026	Smartlead API
03	Visbl visbl.com ↗	Licensed brokers, US · vertical FinTech marketplace	1	3,203	3,203	53	1.65%	22 interested · 41.5% of replies	Jan 2026	Dashboard
04	getdot.ai getdot.ai ↗	RevOps, data and analytics leaders · AI analyst	5	19,677	11,857	94	0.80%	32 positive · 34.0% of replies	Jan – May 2026	Dashboard
05	BuildWithWave buildwithwave.com ↗	Restoration companies, US · vertical AI SaaS	2	5,296	5,296	58	1.76%	9 opportunities · $4,500	Mar – May 2026	Dashboard
	Blended	5 clients · 34 campaign cells	34	54,823	34,685	725	2.09%	151	2026

Reply rate = replies ÷ leads contacted, as the sending platform reports it (BuildWithWave: Instantly’s basis; raw 58 on 5,296)Skydive + Pemo pulled live from the Smartlead API · others from the dashboard screenshots shown on each card

01Message-market fit / Series A / AI teammatesskydive.com ↗

Skydive.

Twenty campaigns. 15,029 emails.
One variable at a time.

Built the outbound estate from zero: sending domains, warmed inboxes, per-domain rate limits, then twenty campaign cells of about five hundred leads each, one variable moving per cell, settings mirrored field by field. The best cell replied at 8.8%, the worst at 0.6%, on the identical product with the identical inboxes. Eight cells were built solo in nine days; the dead ones were killed within a week of launch.

WHAT IT PROVED

The audience beats the angle by a mile. Which audience is the client’s data, not mine to publish. The method is the point: one variable, five hundred leads, kill on the numbers.

HOW IT WAS RUN

Every campaign here was created, configured and monitored through the Smartlead API from Claude Code. Settings mirrored via API, sequences posted via API, stats pulled via API. Nobody clicked through a dashboard, and every number on this page was pulled the same way.

Built with Smartlead · Clay · Prospeo · Apify · Hypertide · InboxKit · Claude Code

See who replied

15,029 sent · 8,204 leads · 398 replies · 1.43% bounce · 0 unsubscribes. Live platform pull, 17 Sep 2026. Interested counts are a floor: most replies sit uncategorised.

Skydive · 20 campaigns · live pull 17 Sep 2026 15,029 emails.
398 replies.
0 unsubscribes.

8,204 leads across twenty cells, 35 marked interested (a floor: most replies sit uncategorised). Best cell 8.8%, worst 0.6%, same product, same inboxes, same window. 1.43% bounce.

Domains · warmup · rate limits · built via API ↗

02Zero to one / Series A / UAE corporate cardspemo.io ↗

Pemo.

No domains, no inboxes, no lists.
Then 25+ meetings a month.

Built the sending estate from scratch: secondary domains, staged warmup, DMARC enforced before the first send went anywhere near the brand. Lists enriched in Clay against UAE-specific sources and screened against the customer base before launch, not after. Six cells, each a single audience against a single vertical, plus one reactivation of closed-lost accounts.

THE LESSON

The winner was the warm list. One closed-lost reactivation out-produced five cold verticals put together. The cheapest pipeline in the building was already in the CRM.

HOW IT WAS RUN

Built with Smartlead · Clay · Hypertide · Claude Code

11,618 sent · 6,125 leads · 122 replies · 53 interested · 0 unsubscribes. Live platform pull, 17 Sep 2026. Tracking off, so opens do not exist as a metric here.

Pemo · 6 campaigns Live pull 17 Sep 2026 11,618 emails.
122 replies.
53 interested.

6,125 leads across six cells. The largest cell, a reactivation of closed-lost accounts, produced 105 of the 122 replies; of the first 69 read by hand, 30 were buying questions.

Per-domain rate limits · stop on reply · plain text ↗

04 / Three more

The list is
the message.

Where the list came from mattered more than what it said.
Click any window to enlarge the platform readout.

Platform readout · Jan to May 2026 ● ● ●getdot.ai / all campaigns↗ getdot.ai campaign dashboard: 19,677 sent, 94 replies, 32 positive

34% positive reply rate

03Cold outbound / AI analytics SaaSgetdot.ai ↗

getdot.ai.

Five angle families ran as separate cells against the same audience of data and analytics leaders: stack detection, recurring exec reporting, business-review prep, a single diagnostic question, the internal request queue. Positive-reply rate was the only verdict. 19,677 sent, 94 replies, 32 positive: 34%. Copy withheld, client work.

Platform readout · registry-sourced list ● ● ●mortgage brokers / NMLS↗ FinTech campaign dashboard: 3,203 sent, 53 replies, 22 interested

41.5% positive reply rate

04Cold outbound / vertical FinTech marketplace / mortgage brokersvisbl.com ↗

Visbl.

Pulled the list straight from the NMLS, the federal registry for the mortgage industry, then enriched and qualified it in Clay on company size, office footprint and active licences. 3,203 sent, 53 replies, 22 interested. Bounces ran high because registry data is not database data, even after a three-step validation pass. Worth it for a list nobody else had.

Platform readout · two signal plays ● ● ●buildwithwave / restoration↗ BuildWithWave campaign dashboard: 5,296 sequences, 58 replies, 9 opportunities

9 opportunities · $4,500

05Cold outbound / vertical AI SaaS / restorationbuildwithwave.com ↗

BuildWithWave.

Two plays for restoration companies. A negative-review detector in Clay and Apify that reads every public review and surfaces the customer complaint this product solves, so the complaint becomes the reason for the email. And a custom Python scraper of the IICRC certification registry, which no enrichment tool sells, to reach only the operators professional enough to have bothered. 5,296 sequences, 58 replies, 9 opportunities. Both plays swap "who fits the firmographics" for "who has shown the problem in public". The second question has far fewer people in the answer.

Watch the automation walkthrough

Who replied

Replies came in
from teams at.

Pooled across all five clients on purpose: no company here is attributable to a client.
490+ companies replied in total; this is a selection.

Sendososendoso.comAffinityaffinity.coHippocratic AIhippocraticai.comConductorconductor.comClaravineclaravine.comAxios HQaxioshq.comClayclay.comProvisprovis.aeBlack Forest Labsblackforestlabs.aiVast.aivast.aiCoefficientcoefficient.ioLabsterlabster.comTimelefttimeleft.comPolyAIpoly-ai.comTamrtamr.comShapoorji Pallonji Propertiesshapoorjiproperties.aeAlfardanalfardan-corporation.comStravastrava.comMaven Clinicmavenclinic.comOMRomr.combeehiivbeehiiv.comFingerprintfingerprint.comDynamic Propertiesdynamicproperties.aeHG Insightshginsights.comAtScaleatscale.comWasabiwasabi.comFunction Healthfunctionhealth.comSendososendoso.comAffinityaffinity.coHippocratic AIhippocraticai.comConductorconductor.comClaravineclaravine.comAxios HQaxioshq.comClayclay.comProvisprovis.aeBlack Forest Labsblackforestlabs.aiVast.aivast.aiCoefficientcoefficient.ioLabsterlabster.comTimelefttimeleft.comPolyAIpoly-ai.comTamrtamr.comShapoorji Pallonji Propertiesshapoorjiproperties.aeAlfardanalfardan-corporation.comStravastrava.comMaven Clinicmavenclinic.comOMRomr.combeehiivbeehiiv.comFingerprintfingerprint.comDynamic Propertiesdynamicproperties.aeHG Insightshginsights.comAtScaleatscale.comWasabiwasabi.comFunction Healthfunctionhealth.com

Replitrepl.itCodeSignalcodesignal.comDigitsdigits.comAllTrailsalltrails.comTradelingtradeling.comDescriptdescript.comLumoslumos.comPlanhatplanhat.comSpotHerospothero.comObserve.aiobserve.aiAirOpsairops.comHarmonicharmonic.aiBoopinboopin.comAllsopp & Allsoppallsoppandallsopp.comPayrailspayrails.comGoldmarkgoldmark.aeOpenWebopenweb.comBridgebridge.xyzGuardzguardz.comReplicantreplicant.aiNuwa Capitalnuwacapital.ioSauce Labssaucelabs.comLocation Solutionslocationsolutions.comAgoraagora.aeTwelve Labstwelvelabs.ioMintlifymintlify.comParityparity.ioReplitrepl.itCodeSignalcodesignal.comDigitsdigits.comAllTrailsalltrails.comTradelingtradeling.comDescriptdescript.comLumoslumos.comPlanhatplanhat.comSpotHerospothero.comObserve.aiobserve.aiAirOpsairops.comHarmonicharmonic.aiBoopinboopin.comAllsopp & Allsoppallsoppandallsopp.comPayrailspayrails.comGoldmarkgoldmark.aeOpenWebopenweb.comBridgebridge.xyzGuardzguardz.comReplicantreplicant.aiNuwa Capitalnuwacapital.ioSauce Labssaucelabs.comLocation Solutionslocationsolutions.comAgoraagora.aeTwelve Labstwelvelabs.ioMintlifymintlify.comParityparity.io

Names and domains from the sending platforms’ reply logs, 17 Sep 2026. No logos, no client mapping, no reply text.

What ties it together

The list is the message.
The audience is the copy.
The data is the boss.

Every figure on this page is a live pull from the sending platform, a hand-classified reply export, or a screenshot of the dashboard. Nothing is inflated, and the one estimate (25+ meetings a month) is the client's own count.

05 / The systems

Things I built
that run without me.

Two Claude Code apps, one list engine,
two n8n pipelines. All in production.

houseofgrowth.vercel.app / dashboard / outset.ai

AI citations84%

Mentions41

Share of voice12%

ChatGPT · Claude · Perplexity · Gemini ↗

01Claude Code build / GEO, AI search

AI Visibility Dashboard.

For Houses of Growth. Enter any domain and get a full AI visibility report: how often the brand is cited by ChatGPT, Claude, Perplexity and Gemini, share of voice against competitors, and concrete GEO recommendations. Next.js, Supabase, OpenAI, Apify; the query fan-out parallelised with BullMQ and Redis.

Live report: outset.ai GitHub

Universe list engine · recall first, precision second

01
Recall sweepProspeo · 16 countries · every plausible firm in the sector
02
Ground-truth promptLabelled set first · 25/25 on the hold-out before touching the universe
03
Relevance passOpenAI Batch API + web search · in-sector only
04
Ledgered loadSource, rows, verification, suppression screen, logged before launch

02Claude Code build / list intelligence

Universe List Engine.

Sector-complete prospect universes for Houses of Growth's finance push: investment banking and M&A advisory across the US, Canada and 15 European countries, then private equity. Sweep everything that could be in-sector, then let a ground-truth-calibrated LLM pass decide what stays. Off-the-shelf databases give you whoever paid to be listed. This gives you the sector.

Brand OS · a brand kit agents can read

Brand voicetone docs

ICP builderideal customers

Positioningmessaging framework

GTM playbooklaunch strategy

Claude API · Next.js · one input, one kit ↗

03Claude Code build / brand intelligence

Brand OS.

Built for Qove.ai. Input a company and get a complete brand kit: voice guidelines, ICP definitions, positioning framework, messaging hierarchy and GTM playbook, structured so content and outreach agents can build on it and stay on-brand without a human checking every output.

Live demo GitHub

n8n · slack → research → claude → drive → sequencer ● ● ●lead-magnet-generator.json↗ n8n workflow canvas for the lead magnet generator

04n8n build / AI automation

Lead magnet generator.

Drop a company domain in Slack. The workflow finds decision makers, maps the competitor landscape, scrapes the company's own site, then has Claude write a lead magnet report specific to that company. Saved to Drive, team pinged, contact pushed into the sequencer. Research to send with zero manual steps.

Read a sample output

n8n · competitor ad audit → live programmatic SEO post ● ● ●trymesha.com / blog / ad-audits↗ Published ad audit page on the client domain

05n8n build / martech / shipped in under a month

Ad audit to programmatic SEO.

Triggered from Slack, the workflow audits a competitor's Facebook ads and landing pages for message alignment and technical performance, then publishes the findings as a live programmatic SEO post on the client's domain. o3 for the reasoning pass, Claude for the human-grade write-up, Apify and Firecrawl covering about 90% of target web assets, and a decision tree that separates high-priority findings from noise before anyone gets alerted. Internal research becomes a public marketing asset with no human in the loop.

View a live output Workflow canvas ↗

06 / The toolkit

What it's
built with.

Production code, not configuration.
Python, TypeScript, SQL, and the GTM stack around them.

OUTBOUND INFRASTRUCTURE
Sending
Smartlead · Instantly · EmailBison · Heyreach · Hypertide · Zapmail · InboxKit
DATA & ENRICHMENT
Lists
Clay · Prospeo · Apify · Firecrawl · Zenrows · Apollo
AUTOMATION & CRM
Plumbing
n8n · make.com · HubSpot · Attio · Customer.io
AI & ENGINEERING
Code
Claude Code · MCP · Cursor · Next.js · TypeScript · Python · SQL · Supabase / Postgres

07 / Next

Let's build something
that runs itself.

Open to GTM engineering roles and fractional engagements. I ship the infrastructure, the data, the signals and the agents, then hand your team a system it can run. Based in Mumbai, working globally.

01vaibhavva219@gmail.com 02linkedin.com/in/vaibhavvaher 03github.com/VaibhavAher219 04x.com/vaibhavvaa Resume, one page (PDF)

Replies within a day, India hours. No agency, no subcontracting.

Outbound,engineered.

Solving revenue problems,one agentic loop at a time.

Production code,tied to revenue.

Otera

BetterFutureLabs

Ateliex

Four clients

Real plays.Real numbers.

Skydive.

Pemo.

The list isthe message.

getdot.ai.

Visbl.

BuildWithWave.

Replies came infrom teams at.

Things I builtthat run without me.