AI Labs Engineer

AI Labs exists to make AI useful in the day-to-day work of Claspmates. Over the past year, they’ve built internal products and workflows, helped teams develop practical AI skills, and created shared systems they can reuse. Today, over 70% of Claspmates use an agent/tool AI Labs built every week—and they’re ready to build on that momentum.

We’re hiring an early-career engineer as the second person in AI Labs to help shape what comes next. You’ll work across the full product loop: understand a workflow, identify the right problem, build across the stack, launch, see how people use what you built, and improve it.

This is an L1 role where you’ll get autonomy, feedback, and support while learning to own meaningful problems from beginning to end. You won’t spend your time implementing perfectly specified solutions; instead, you’ll help figure out what should be built and why.

What you’ll do:

• Own problems from discovery through adoption. Work directly with internal users to understand their workflows, build and launch solutions across the stack, and measure whether they helped.

• Build and improve internal tools. Refine existing agents and create new ones, including their instructions, tools, guardrails, evaluations, and human handoffs.

• Make your work maintainable. Document decisions, setup, risks, and operating steps. Raise architecture, security, privacy, compliance, or production risks early.

• Help Claspmates build with AI. Run programs such as AI Champs, Lattes with AI Labs, and Lunch & Learns, and pair with colleagues to turn ideas into shipped products.

• Keep learning as the field moves. Explore new models, products, and agent patterns, follow what is happening across the industry, and test what is genuinely useful.

What we’re looking for:

• Evidence you can ship. You have built and shipped at least one tool/product/project beyond your own machine.

• Practical AI fluency. You actively use AI models and development tools. You can explain where they help, where they fail, what you verified yourself, and how you would evaluate quality in a real workflow.

• Full-stack range. You can reason about interfaces, backend logic, APIs and integrations, data flows, testing, and deployment. You do not need equal depth everywhere, but you can own a bounded feature end to end.

• Product judgment. You talk to users, clarify the decision the product should enable, choose the smallest useful version, and measure whether it worked.

• Fast learning. You ask for help with useful context, can operate with incomplete information, want coaching, and apply it.

You do not need to check every box above to apply. If you bring high agency, are ready to dive into unfamiliar problems, and are excited to learn as you go, we want to hear from you.

Interview process:

The process is designed to understand how you build, think, learn, and collaborate—not to test trivia.

1. Hiring Manager call. A conversation with Harsh about completed work, product judgment, learning, and how you operate.

2. Technical assessment. A realistic, bounded engineering pairing session. AI tools are allowed; they care how you reason, build, debug, and verify.

3. Case study. A conversation about product thinking, scope, tradeoffs, communication, and taking an AI product from idea to adoption.