About Pritish Budhiraja

At TestZeus I own the frontend architecture end-to-end: feature-sliced modules, the Vite build, Docker, and Kubernetes on GKE. The job is making the codebase something a team can keep changing without it falling apart — including repo-wide formatting and structure with Biome, so reviews stay about the change, not the style.

A large part of that work is the Teach Agent. You record a browser session; the platform turns it into Gherkin scenarios and executable Python tests. That only works if artifact processing is asynchronous and generation is progressive — not a single prompt dumped into a box. I also built the Composio-based integration layer so connecting GitHub, Jira, or another source of truth is an extension, not a rewrite. Fifteen-plus apps sit on that path today: connect an account, build a knowledge base, generate tests.

On the commercial side I implemented usage-based billing and quota enforcement with Flexprice and Redis — real-time events, per-user metrics, cancelled-run recovery, and clear feedback when a quota is exhausted. Security and reliability sit next to that: JWT revocation, tenant-derived auth, login audit trails, deployment health checks, and Sentry.

I joined Juspay as an Associate Software Development Engineer in late 2022 and left as an SDE I in 2025. Most of that time was Hyperswitch.

I led the Hyperswitch Dashboard from zero to production in ReScript and React, so merchants could configure 25+ payment processors without a backend change for common cases. I also built HyperStudio for layout and white-labeling, and owned the Web SDK across six merchants: dynamic fields per payment method, faster first load via tree-shaking and lazy-loading, and CloudFront Functions for version-pinned rollouts. CI/CD went through Jenkins to S3 and CloudFront, with GitHub Actions as PR gates.

The Web SDK is open source. I triaged production defects from merchants on that repo, and I mentored five engineers through design reviews and onboarding.

Before Juspay I shipped seller-side purchase journeys at ZFunds — health insurance and fixed deposits — and interned at HighRadius on ML-assisted billing and collections: Spring on the back end, React dashboards, and Python scoring services.

The work I like most sits at the boundary. React and TypeScript for the product. AWS, Docker, and Kubernetes for how it ships. Redis when usage or rate limits have to be correct, not approximate. I care about the boring parts: lint and format consistency, PR checks that catch regressions, and rollouts that don't surprise merchants.

On the side I've built a Fastify URL shortener with Redis analytics, and an API gateway that implements token bucket, sliding window, and fixed window in Redis — with comments on every command, because I wanted the code to teach the algorithm, not just run it.