Primary record

AI Builder Intern

Scale AI Indexed employerSan Francisco, CA · San Francisco, California, United States
Source-hosted applyChecked 3h agoInternship
Apply at Scale AI

Scale AI receives this application through Greenhouse. Babu Careers does not claim delivery.

Workplace

On-site

Employment

Internship

Published

Aug 11, 2026

Closes

No date supplied

The role

ABOUT THE ROLE

This isn't a research internship. You won't spend the summer writing reports or sitting in strategy meetings. You'll spend it building — shipping AI-powered tools, automating workflows, and deploying agentic systems that real teams at Scale AI use every day.

Embedded in the Data & Technology org, you'll work directly alongside engineers, data scientists, and ops leads on live automation initiatives. If you have strong instincts for what AI can do today, a bias for building over theorizing, and a fluency in modern LLM tooling — this role is for you.

────────────────────────────────────────

WHAT YOU'LL BUILD

Agentic Workflows & Automation

→ Design and deploy multi-step agentic workflows using LLM-integrated frameworks (LangChain, LangGraph, CrewAI, or similar)

→ Build API-connected automations that tie together internal tools — Slack, Salesforce, Notion, and internal data systems

→ Prototype and iterate fast; build things that work, then make them better

AI Tooling & Internal Products

→ Develop lightweight internal tools and dashboards that surface AI outputs to business teams

→ Vibe-code functional UIs — React, plain JS, or whatever gets to working fastest — for internal adoption

→ Identify friction points in current workflows and propose AI-first replacements

Measurement & Signal

→ Instrument your own work — capture usage signals, time-saved estimates, and adoption metrics from day one

→ Contribute to the org's ROI measurement framework by tagging your projects to defined value categories

────────────────────────────────────────

A NOTE ON "VIBE CODING"

We're not precious about how code gets written. Use Cursor, Claude, Copilot, or whatever gets you from idea to working demo in hours, not weeks. We care about judgment, craft, and what ships — not keystroke attribution.

──────────────────────────────────

Requirements

Department: Data, Technology & Systems