Nilofer Rajpurkar

Nilofer Rajpurkar develops tools that bring financial infrastructure into the environments where developers and AI agents build applications. As Stripe’s Product Lead for Agents and Developer Experience in its 2026 developer keynote, her work included payment-integration guidance, sandbox provisioning, and approval rules for agents taking financial actions. These address different parts of the same practical problem: helping an agent build a working payment integration and defining which actions it can perform afterward.

From mobile builds to developer platforms

Nilofer joined Microsoft after earning a bachelor’s degree in biomedical engineering at Purdue University. As a program manager for Visual Studio App Center, she worked on mobile developer tools. Her writing on hosted macOS build environments addressed a compatibility problem for mobile teams: gaining access to newer Apple tooling while retaining environments needed to build and test older projects.

In a 2018 article about App Center’s GitHub integration, she explained how developers could move from a mobile-project pull request to configuring continuous integration and receiving build results inside GitHub. The integration brought build feedback into the place where developers were already reviewing code.

By 2021, her work included GitHub Packages. She authored the general-availability announcement for GitHub’s Container registry, explaining organizational ownership, fine-grained permissions, and publishing through GitHub Actions. Those capabilities let teams distribute container images alongside their code while managing image access independently of a repository’s visibility.

Her later Stripe work extends this attention to developer workflows into agent-assisted application building. In Inside the AI Economy, she demonstrated adding payments to an application in Claude using Stripe’s MCP server and agent skills, including a sandbox provisioned through the CLI. The example connects implementation guidance with an environment in which the resulting integration can be tested.

Guiding integrations and governing financial actions

Nilofer’s Stripe examples distinguish several decisions that payment-capable agents must handle:

• Choosing an integration for the business: Her developer-keynote examples connect implementation choices to customer location, checkout requirements, and tax considerations. An agent needs those business details to choose an appropriate integration; producing plausible payment code does not answer them.

• Testing before going live: Sandboxes provide an environment for developing and testing payment integrations. Her Claude demonstration includes provisioning that environment from the CLI, making setup part of the agent-assisted building process.

• Escalating financial actions: Her agent-operated customer-support examples use explicit refund approval thresholds. A smaller refund can proceed automatically, while a larger request returns a structured response directing the agent to seek human approval. The financial system enforces the threshold and tells the agent what must happen next.

Building for agents as customers

Nilofer also examines how AI changes the businesses using these tools. In Inside the AI Economy, she uses Stripe data to discuss faster development, international sales from the outset, changing pricing models, and new sales channels. Replit’s combination of subscription and usage pricing illustrates how an AI business can combine recurring access with charges tied to consumption.

Her discussion of agents as buyers extends the developer-experience question beyond code generation. She reports that agent traffic to Stripe’s documentation grew tenfold in a year and argues that companies need to adapt their product and revenue infrastructure to this audience. Documentation, integration guidance, testing environments, and financial permissions become parts of how software agents discover, build with, and act through a service.