Agent Systems Engineer is what we call this role. You may know it as GTM Engineer, Revenue Systems Engineer, Growth Engineer, or internal Forward Deployed Engineer. The substance is the same. You’re a deeply technical operator who embeds with revenue teams, learns how the business actually moves, and rebuilds workflows as governed, AI-augmented systems
You write SQL fluently. You live in APIs. You’ve shipped production GTM automation that other people depend on. You know what good revenue data architecture looks like (warehouse as source of truth, modeled in dbt, activated into Salesforce and the rest of the stack via Reverse ETL) and you have opinions about why most companies get it wrong
You also believe in practical AI. You’ve deployed LLMs against real GTM problems where the business value was concrete (account research, classification, enrichment, content generation), and you have honest views about what worked, what didn’t, and where the hype outpaces reality
Most companies treating AI as a productivity tool are pointing it at individual jobs. We think the bigger opportunity is rebuilding entire revenue processes around agents and modern data infrastructure. We sell that thesis to product security teams every day. Running our own GTM motion the same way is how we hold ourselves to the standard we’re selling
You’d be first in seat at Finite State, so you own the full motion: discovery with revenue leaders, system design, build, governance, evaluation, rollout, and the runbook so it survives you
Run discovery with revenue leaders before building. Sit in pipeline reviews. Watch a deal cycle end-to-end. Find the actual time sinks before designing a solution
Architect the GTM data layer: Snowflake (or equivalent) as source of truth, dbt for modeling, Reverse ETL (Hightouch, Census) for activation into Salesforce, HubSpot, Outreach, and the rest of the stack
Design and build AI agents and AI-augmented workflows for revenue-critical work: account research, ICP scoring, signal-based plays, outbound personalization, CRM enrichment, deal intelligence, churn risk, expansion triggers, lead routing
Deploy LLMs and agents where they add real business value, and skip them where they don’t. We’re not interested in AI for the sake of AI
Wire agents and systems together via APIs, webhooks, MCP servers, and lightweight code (Python, SQL, TypeScript). Use platforms like Clay, n8n, Workato, or Hightouch AI when they fit. Build custom when they don’t
Build signal pipelines that capture buying intent (hiring patterns, funding events, security disclosures, product telemetry from our own platform) and trigger the right agent or action automatically
Stand up the governance layer for every agent you ship: permissions, audit trails, access controls, sensitive data handling, and rollback paths
Build evaluation harnesses that measure real business outcomes (pipeline generated, deals accelerated, rep hours saved), not just whether the agent ran
Codify recurring patterns as reusable skills so the next agent doesn’t start from scratch
Document the architecture and write the runbook so the next person on the team can learn from your work
Expand into adjacent functions (Finance, People, Security ops) as the pattern proves out