AI-Native Software Engineer, QA Automation - Zenara Health

About Zenara Health

Zenara Health is a technology-oriented mental healthcare organization building clinical software for behavioral-health clinics. Our products, Assess and Zenflow, support assessments, care management, clinical workflows, and operational delivery. We are a startup building an AI-native engineering team where quality must scale at the same speed as development.

The Opportunity: AI-Native Software Engineer, QA Automation

We are seeking a Mid to Senior AI-Native Software Engineer focused on QA Automation. This is a software engineering role dedicated to building systems that prove product functionality, not manual testing. Our developers leverage AI agents, leading to rapid code generation. We need a QA automation engineer who can build the testing harness, automated checks, synthetic data, test environments, and agent-driven QA workflows to keep pace with agent-speed development. We are looking for someone who demands evidence and can build robust test systems.

Our AI-Native QA Vision

At Zenara, QA is an engineering platform, not a downstream gate. We are building a system where product requirements transform into test plans, test data, automated browser and API checks, evidence reports, failure routing, and release confidence. AI agents will assist in creating and running this work, with engineers governing quality.

You will help design and operate this comprehensive QA system, including:

• Browser and API test runners
• Synthetic test data
• Isolated test environments
• Requirement-to-test workflows
• Automated checks on every change
• Agent-driven QA workflows
• Evidence generation for releases
• Flaky-test reduction
• Failure routing and debugging
• Quality metrics and release confidence

Our goal is to build a system that catches real defects proactively.

What You’ll Do

• Build and maintain the QA automation harness across browser, API, data, and workflow layers.
• Create automated checks that run on every meaningful change.
• Build synthetic test data and isolated test environments.
• Own the UAT/QA automation pipeline from requirement to test plan to run to evidence.
• Use AI agents to help generate, execute, improve, and debug tests.
• Build independent agent-driven QA workflows that test the work of code-writing agents.
• Prove tests work by reintroducing known bugs and showing they fail correctly.
• Reduce flaky tests and distinguish real product failures from environment noise.
• Produce reliable QA evidence for releases.
• Work with engineering and product to define quality gates for critical user journeys.

What We’re Looking For

• Strong software engineering ability: design, write, debug, and maintain code.
• Experience with test automation using Playwright, Cypress, API testing frameworks, or similar tools.
• CI/CD experience and comfort wiring tests into build and release workflows.
• Practical experience using AI agents to create or improve tests.
• Strong judgment about whether a test actually proves anything.
• Ability to seed a bug, run a test, and demonstrate that the test catches the failure.
• Comfort debugging flaky tests, environment issues, and test data problems.
• Skeptical, evidence-driven thinking.
• Clear communication around risk, coverage, and release confidence.

Our Current Stack

• TypeScript
• Node.js, Hono
• React
• Prisma, PostgreSQL
• Redis, BullMQ
• Playwright
• Azure AKS, Blob Storage, Azure Monitor / OpenTelemetry
• Terraform, Helm, CI pipelines
• Medplum / FHIR
• Flowable
• Claude Code, Codex
• Dify-based AI workflows

What “AI-Native QA” Means for This Role

For this role, being AI-native means you can:

• Use agents to generate test ideas, test code, and edge cases.
• Review agent-created tests critically.
• Build workflows where agents test independently from the agents that wrote the code.
• Use AI to analyze failures and propose debugging paths.
• Create repeatable quality gates that scale with development speed.
• Measure whether AI-generated tests are actually useful.
• Avoid false confidence from tests that pass without proving behavior.

The objective is trustworthy evidence, not just more tests.

Nice to Have

• Healthcare or regulated-software experience.
• Experience testing AI-powered features, including output quality.
• Experience building test environments and synthetic data.
• Performance or load testing experience.

Working Hours & Commitment

This is a full-time role with an approximate commitment of 50 hours per week. All meetings and live collaboration happen between approximately 6:00 PM and 12:00 AM IST, overlapping US Pacific Time. Attendance in that window is required; your remaining work is yours to schedule.

What Success Looks Like

• First 30 days: Map the current QA setup, agree on highest-risk user journeys, and deliver one automated end-to-end QA lane that runs on every change and catches a known bug.
• Within 90 days: Key journeys for Assess and Zenflow will be automatically covered, every release will have QA evidence, and flaky-test rate and run time will be measured and improving.
• Over time: Build a mature agentic QA platform where escaped defects are low and QA throughput keeps pace with agent-speed development.

Compensation and Benefits

• Competitive compensation, based on demonstrated ability and experience.
• Fully remote work.
• Equipment allowance.
• Direct access to the founder and Head of Engineering.
• The opportunity to build the QA function as an engineering platform from the ground up.

Who Will Thrive in This Role

This role is ideal for someone who:

• Thinks like an engineer, not a manual checklist executor.
• Enjoys building tools, systems, and automation.
• Is naturally skeptical and evidence-driven.
• Cares about proving behavior, not just producing green checkmarks.
• Wants to make QA scale with AI-native development.
• Likes working in fast-moving, ambiguous startup environments.

You may struggle if you only have manual testing experience, cannot code, sign off work without evidence, or write tests that pass regardless of whether the product is correct.