Piyush Pathak. Staff-level Software Engineer in Test with 13+ years across SaaS, distributed systems and enterprise platforms. I own quality end to end, then roll up my sleeves on the full stack: TypeScript, Node, Next.js, AWS serverless and Salesforce.
I started in test engineering, building automation frameworks when most teams treated QA as an afterthought. That foundation shaped how I build: quality is a property of the system, not a phase at the end.
Today I work at a staff level — owning test strategy across UI, API and backend, designing automation frameworks that stay fast and non-flaky at scale, and wiring quality gates into CI/CD so broken code never reaches a customer. On the platform I helped take coverage from 60% to ~90% with gated deployments.
What I enjoy most is that I don't stop at the test layer. I ship the product too — Next.js front ends, serverless APIs on AWS, Salesforce backends — which means I design for testability from the inside out, not bolt it on afterwards.
Lately I've been folding AI and LLMs into the testing loop: AI-assisted test generation, code review and evaluation workflows built on Claude Code, the Anthropic API and MCP.
Test engineer → test lead → full-stack SDET. The same instinct throughout: make systems provably reliable.
I'm hired as a test architect — but I'm dangerous because I also build the thing. Both columns are day-to-day work, not a wishlist.
Led migration of 14M+ files from Dropbox to an AWS serverless platform integrated with Salesforce, building agents that detect and repair problematic files in real time so customers never saw the gaps.
Helped deliver an enterprise pharma application: Salesforce backend, serverless API layer, Next.js front ends — installable into any org via a managed package. Built the testing strategy alongside the features.
Designed an in-house automation framework that replaced live network calls with mocks/stubs — cutting infrastructure cost and slashing flaky failures across Shutterstock's web platforms.
Built an everyday workflow that uses Claude Code, the Anthropic API and MCP for AI-assisted test generation, code review and evaluation — accelerating coverage on complex distributed systems.
A Claude agent that authors, generates, triages and self-heals Playwright tests for any web app — driven from Claude Desktop as an MCP server or from CI as a CLI, with the whole loop running as a deployment gate in GitHub Actions.
Compiles plain-English intent into a structured test plan.
Explores the app from a URL and writes real Playwright specs with assertions.
Classifies each failure — real bug vs DOM drift vs flake — with a root-cause report.
Rewrites the broken locator, verifies green, and opens a fix PR.
Beyond using AI to write tests, I'm building toward evaluating AI itself — prompt validation, output consistency and automated testing workflows. As products ship LLM features, someone has to make them provably reliable. That's the same job I've always done, with a new surface area.
Open to staff / principal SDET and full-stack engineering roles, remote-first. The fastest way to reach me is email.