Menu

Systems I've built

A performance marketer who became technical because the numbers he was handed weren't good enough. These are the systems that came out of it, each in daily use.

Lead pipeline and attribution

Ayurvedic wellness group (Kairali)
  1. Meta Lead Ads
  2. Google Sheets
  3. Apps Script
  4. Supabase
  5. Attribution join
What I built
Meta Lead Ads → Google Sheets → Apps Script (time-driven triggers, validation on write) → Supabase Postgres, with identity resolution and an attribution join. Zero third-party connectors.
What it changed
Shows exactly where the funnel breaks: creative fatigue, wasted spend, negative-keyword gaps, and which regions send leads that actually convert.
Meta Graph APIApps ScriptSupabaseSQL

Marketing intelligence dashboards

Ayurvedic wellness group (Kairali)
  1. Meta Ads
  2. Google Ads
  3. GA4
  4. GSC
  5. Master view
What I built
Dashboards for Meta Ads, Google Ads, GA4, Search Console, content QA, content generation, and a master view joining paid, organic and lead data.
What it changed
One place to see spend, CTR, conversion and organic trends side by side, so budget and content decisions come from joined data instead of separate platform reports.
Next.jsTypeScriptSupabaseRechartsVercel

Multi-entity CRM

Ayurvedic wellness group (Kairali)
  1. New lead
  2. AI quality check
  3. Sales team
  4. Conversion / cold
What I built
CRM across multiple business entities, tracking every lead from capture to outcome.
What it changed
Shows where each lead is stuck (sales team or AI quality check), how the conversion flow is moving, and why cold leads went cold, with recorded reasons.
Next.jsTypeScriptSupabase

AI inbox triage agent

Personal workflow
  1. Inbox
  2. Classify
  3. Draft reply
  4. Flag to me
What I built
Agent that classifies incoming email and drafts replies. Permissions expand step by step; it never acts beyond what it has been cleared to do.
What it changed
Daily briefing of what matters, replies drafted for routine mail, and anything important or unclear flagged to me instead of handled blindly.
Anthropic APIGmail API

AI voice lead qualification

Ayurvedic wellness group (Kairali)
  1. Inbound lead
  2. AI call
  3. Intent score
  4. Human confirms route
What I built
Voice agent that calls inbound leads, classifies intent, and proposes routing that a human confirms.
What it changed
Lead quality is assessed on the first touch, so sales time goes to high-intent leads and low-quality sources show up early.
ElevenLabsAnthropic APISupabase

AI content + QA system

Ayurvedic wellness group (Kairali)
  1. Brief
  2. Draft
  3. Compliance gate
  4. AI judge score
  5. Publish
What I built
Content generation with compliance pass/fail gates and LLM-as-judge scoring before anything is published.
What it changed
Faster content output, plus ready-to-use scripts for ad creatives, with compliance checked before anything ships.
Anthropic APIOpenRouter

Content compliance checker

Personal brand content
  1. New content
  2. Rule checks
  3. CI check
  4. Pass / block
What I built
Node.js checker with enforced patterns across multiple categories and test fixtures, run in GitHub Actions on every change, plus a proof ledger that blocks previously rejected claims.
What it changed
Fewer compliance flags on published content, because issues are caught automatically before anything goes out.
Node.jsGitHub Actions

Want to see one in detail?

I'm happy to walk through how any of these works.

Message me on LinkedIn