Selected systems

In progress: an AI Ops engagement with the marketing organisation. I'm building their internal AI layer — agents, connectors, and signal pipelines — across a prioritized set of eleven use cases, then handing the whole thing over so the team runs and extends it without me. Four months, from architecture to internal autonomy.

A from-scratch account-based engine: it ingests inbound and lead-gen signals, scores them against a custom ICP model, and enriches every contact through a multi-provider waterfall. The hard part was a CRM sync that matches companies by domain across messy parent/subsidiary structures and writes back to Salesforce cleanly — running unattended on scheduled jobs across a 15,000-account market.

A full account-based scoring system that turns raw website-intent signals into ranked, sales-ready accounts. Custom backend functions auto-enrich missing emails through a private API and route the best accounts straight to the BDR team, with audit logging end to end — so reps open a list that's already prioritized.

A fully autonomous revenue engine on a platform the client owns: it scrapes engagement signals, qualifies prospects with AI, scores them across multiple dimensions, and runs both acquisition and retention outreach on its own — catching at-risk members and firing win-back automatically. Wired to live Stripe and finance data.

A continuously-running pipeline that scores, enriches, and routes inbound leads in near real-time and pushes them to the CRM on a short cron, deduping against the existing customer base automatically. It also generates and feeds multi-country outbound sequences.

A centralized intelligence platform (348 companies, 3,150+ contacts) with a live signal-scoring model and a learning engine that turns the team's own corrections into reusable scoring rules. Includes a paginated CRM integration engineered around the source API's hard limits.

A real-time intent-detection pipeline with multi-source scraping, AI qualification, and automated outbound. The mission was a migration: I rebuilt the entire GTM stack off rented tooling onto infrastructure the client fully owns — at feature parity, in two weeks.

A self-hosted, AI-powered signal-scoring API running on the client's own infrastructure, with no per-seat tooling cost. It cross-references advertising engagement with firmographic fit to surface the hottest accounts for sales.

I took their entire CRM and built scripts to refresh the data they had — cleaning, deduplicating, and re-enriching existing records. Then I built a full scoring machine that runs on top of it: automated scoring on new lead ingestion, plus an activation system for the Sales team.