Responsibilities

• The accounts. Full builds to our written account standard — and Editor CSVs that import clean the first time. Campaigns, ad groups, match-type discipline, RSAs with a pinning strategy, the full asset set at the right level, negatives and their associations, audiences with observation vs. targeting set deliberately, geo with presence-only explicit. Search term mining, PMax asset and listing groups, feeds, pacing.
• The Measurement — Auto-tagging integrity across every account. Conversion action definitions that count leads, not clicks. GTM container integrity — we've had one wiped off a live site and found a client site re-tagged to a foreign GA4 property we still can't account for. Consent mode, enhanced conversions, and the call-tracking → client mapping that determines whether a lead can be attributed at all.
• The Data Stack — ClickHouse and BigQuery Backfills, schema changes, rollups, dimension integrity. The SQL behind every report and health check. Fixing the pipeline when a source changes shape.
• The Automation, On A Short Leash — A weekday monitor for spend spikes, zero-lead days, and tracking breaks — read-only, writes its findings to the client folder, emits the exact command and waits for a human. A monthly report drafter, piloted on two clients and killed if a reviewed draft isn't meaningfully faster than writing from scratch. Extending the MCP server without routing around its safety layer.
• The reporting — TL;DR first: what happened, why, what to do. Cross-client rollups. Which accounts subsidize which. We report real metrics and our system is built to perform so we can show real metrics. We are AdTech and we operate at a different level than any other marketing agency. That's why we don't have competition in our niche of Pool Builder agencies. Our ML snags the leads that buy pools while other agencies over bid on the scraps we don't want anyway. They just can't compete on our level

Requirements

• 4–7 years in paid search or marketing ops, including direct ownership of Google Ads accounts at $50k+/mo aggregate spend, plus real engineering work you can walk us through
• Agency background preferred
• Strong SQL and fluent Python a must
• Google Ads at depth. Smart Bidding mechanics — what tCPA/tROAS actually optimize against, what changing a conversion action does to a live model, where portfolio strategies leak
• PMax and AI Max: signals as signals, brand exclusions, the search-terms visibility ceiling, and catching Search cannibalization in the warehouse instead of arguing about it
• Editor at a professional level: the CSV schema per entity, dependency rules, how partial imports fail, and how to diff a proposed build against a live account before you push it
• Scripts for account checks, the API and GAQL for anything scheduled
• Measurement, properly. GA4's data model and BigQuery export. GTM, dataLayer, consent mode v2, server-side concepts. gclid/wbraid/gbraid, cross-domain, enhanced conversions, offline conversion import — and the honest difference between a modeled conversion and a counted lead. Call tracking, DNI, dedupe
• Data engineering. Strong SQL, ClickHouse preferred. Python for pipelines, API clients, and small tested tools. Idempotent backfills, dedupe, late-arriving data. Real git fluency — branches, reviewable commits, GitHub Actions
• AI tooling with judgment. Claude Code and agentic workflows as working tools, plus a clear sense of where not to point them. Our written position: no orchestrators, nothing that contacts a client, no autonomous campaign changes. Having written or extended an MCP server is a strong plus

Preferred

• Having written or extended an MCP server is a strong plus
• Local-services lead gen is a plus

Benefits

• 401(k) match, full medical/dental/vision, unlimited PTO
• Unlimited PTO with a 15-day floor — we track load and name coverage before anyone goes, so the policy is usable
• Hardware and tooling budget, Claude Max seat included
• Fully remote, async-first; the entire operating system is a git repo, there is no office to be absent from