AI GTM Engineer

About the role

Samba TV is building an AI-native revenue organization from scratch. This role is responsible for closing the gap between the current state of our GTM systems and the potential they hold.

Responsibilities

• Design and operate multi-agent systems connecting Salesforce, Gong, Slack, Snowflake, and Samba's proprietary TV data assets into unified, real-time GTM workflows
• Build and maintain MCP (Model Context Protocol) servers that give Claude governed, real-time access to our GTM stack, including live Salesforce records, Gong call data, and Samba viewership signals
• Eliminate manual data transfer and context-switching for sellers by automating the aggregation, summarization, and routing of deal intelligence across tools
• Create RAG pipelines and prompt libraries that give Claude accurate, governed context on our products, prospects, pricing, and competitive landscape
• Architect integrations between AI services and internal systems using Python and APIs, enriching contact and account records with live signals
• Evals, reliability, and continuous improvement: Design and run structured evals across all production AI systems; measure output quality, accuracy, regression risk, and real business impact, not just vibes
• Build eval frameworks that catch prompt drift, model behavior changes, and degraded tool use before they hit sellers in production
• Run A/B experiments across workflows to prove what is actually moving pipeline and revenue, not just what looks good in demos
• Maintain the feedback loop: what broke, why, and what the fix is
• Own the internal authority on Claude across the Revenue Org: best practices, MCP architecture, prompt engineering standards, and enablement
• Write documentation and run enablement sessions so sellers and operators extract real value from every system you ship
• Signal integration across the sales org: Build systems that automatically surface deal risk, flag renewal exposure, draft personalized outreach, and score accounts by conversion likelihood using real viewership signals
• Design workflow automation that reduces the cognitive load on sellers so they spend more time selling and less time updating CRM fields, hunting for context, or writing the same email for the 40th time
• Identify the highest-friction points in our GTM process and systematically eliminate them through automation

Requirements

3 to 5 years in software engineering, data engineering, or GTM/revenue operations engineering with serious technical depth

Production-level Python and/or JavaScript; you write, ship, and maintain code others depend on

Hands-on LLM production experience: prompt engineering, tool use, multi-turn agentic workflows, RAG

Real eval practice: you know how to measure whether an AI system is working, build regression tests, and catch drift before it causes problems

Deep, production-level Claude expertise including the API, structured outputs, tool use, agentic workflows, and system prompt design; you can also teach it and build reliable systems around it

Hands-on experience with Salesforce (SOQL, APIs, Flows) and Gong

Snowflake and dbt working knowledge for querying and transforming data to support AI workflows

Comfortable in both a command line and a boardroom; you move between implementation and executive conversation without losing either audience