About the Product

We are building an AI-powered test automation platform that enables engineering teams to automatically generate, execute, and maintain end-to-end browser test suites — without writing test code manually. The platform uses large language models to understand web applications, plan meaningful test scenarios, and produce self-healing tests that adapt when the application changes.

The product is already validated against real enterprise applications and is now scaling toward becoming a universal QA automation tool capable of testing any web application with minimal human intervention.

About the Role

We're looking for a Senior AI Engineer to own and advance the AI-driven core of the platform — the prompt pipelines, code generation quality, self-healing intelligence, and the emerging Human-in-the-Loop (HITL) feedback system. This is a hands-on, high-ownership role where your work directly determines the product's capability ceiling.

You will be the person responsible for pushing the system's autonomous pass rate toward 95%+ and making the platform reliably handle applications it has never seen before.

What You'll OwnAI Pipeline Architecture

- Own the end-to-end AI orchestration: from application analysis through test planning, code generation, and self-healing.
- Design the context assembly strategy — deciding what evidence gets fed to the model at each stage and how to stay within token budgets.
- Architect and maintain structured output parsing — ensuring LLM-generated test code is syntactically valid, follows best practices, and uses robust selectors.

Prompt Engineering & Code Generation Quality

- Design, iterate, and benchmark the prompt chains that produce Playwright test specs — covering positive paths, negative paths, edge cases, and destructive operations.
- Build evaluation harnesses to measure generated code quality: selector accuracy, assertion correctness, flakiness rates, and coverage breadth.
- Reduce AI hallucinations — eliminate invented form values, guessed selectors, fabricated success messages, and speculative navigation paths.
- Implement prompt-level guardrails: anti-pattern linters, role-based accessibility locator enforcement (getByRole, getByLabel), and universal widget interaction patterns (comboboxes, date pickers, modals).

Self-Healing & Error Recovery System

- Advance the self-healing engine that intercepts test failures, captures runtime context, and uses AI to produce targeted fixes.
- Build smarter healing strategies: distinguish between genuine app bugs vs. test fragility, prevent assertion weakening during heal cycles, and implement cost caps on retry budgets.
- Design the feedback loop so that healed tests are surfaced for human review rather than silently accepted.

Human-in-the-Loop (HITL) System

- Build the structured data injection pipeline — replacing AI-invented test data with human-supplied values (form fields, dropdown options, expected messages, file fixtures,



date formats).
- Design the App Memory system: a persistent, project-scoped knowledge base that accumulates human corrections and contextual data across test runs, so the AI gets smarter over time.
- Implement interactive review and correction workflows at each pipeline stage (journey planning, spec generation, post-execution triage).

Dynamic Entity Discovery & Context Enrichment

- Build a live entity extraction system: during authenticated crawls, automatically discover available dropdown values, table rows, navigation labels, and form field constraints.
- Design data schemas and injection pipelines so the AI generates tests using known-valid data rather than speculation.
- Integrate API spec parsing (OpenAPI/Swagger) for richer test generation context.

Tech StackLayerTechnologyAI / LLMLLM APIs (commercial frontier models), structured prompting, tool use, streamingBackendPython 3.11+, FastAPI, SQLAlchemy (async), PydanticBrowser AutomationPlaywright (TypeScript)FrontendReact 18, TypeScript, ViteInfrastructureCloud (Azure / AWS), GitHub ActionsRequired Experience

- Applied LLM/AI Engineering — 3+ years of production experience building systems on top of large language models (OpenAI, Anthropic, or equivalent). You understand token economics, context window management, structured output parsing, prompt chaining, and tool use.
- Python — 4+ years professional backend development; strong async Python, REST API design, and data pipeline experience.
- Prompt Engineering at Scale — demonstrated ability to design, benchmark, and iterate on prompt pipelines that produce reliable structured output (code, JSON, plans). Not just chat prompts — production prompt chains with evaluation metrics.
- Code Generation & Program Synthesis — experience with AI systems that generate executable code. Understanding of how to validate, sandbox, and evaluate generated programs.
- Browser Automation / Testing — working knowledge of Playwright, Cypress, or Selenium. You should understand DOM structure, CSS/XPath selectors, accessibility trees, and why browser tests break.
- System Design — ability to architect multi-stage AI pipelines with retry logic, error recovery, cost controls, and human-in-the-loop checkpoints.

Strongly Preferred

- Experience building self-healing or auto-remediation systems — in testing, infrastructure, or any domain where an AI agent diagnoses failures and applies fixes autonomously.
- Background in QA engineering, SDET,



or test infrastructure — you've seen the pain of flaky tests and brittle selectors from the practitioner's side.
- Experience with RAG (Retrieval-Augmented Generation) or persistent memory systems for AI agents.
- Familiarity with AST parsing, static analysis, or compiler tooling — understanding how to extract structure from source code programmatically.
- Experience with AI evaluation and benchmarking — building scoring rubrics, regression test suites for prompts, and A/B testing prompt variants.
- Prior work on developer tools, observability platforms, or CI/CD systems.
- Familiarity with TypeScript and the Playwright test runner API.
- Experience with multi-modal AI — using screenshots or visual context alongside text for improved AI reasoning.

What We're NOT Looking For

- Pure ML/data science researchers — this is an applied engineering role, not a model training role.
- Prompt engineering hobbyists — we need someone who has shipped production AI systems with real users, error handling, and cost constraints.
- Candidates unfamiliar with software testing concepts — the domain context matters.

How We Work

- Small team, extreme ownership — you'll shape the AI architecture, not just implement tickets.
- Ship and measure — we validate changes against real applications and track pass rate improvements quantitatively.
- AI-native workflow — we use AI tooling throughout our own development process.
- Async-first, remote-friendly — no unnecessary meetings, no rigid hours.
- Direct product impact — the AI pipeline is the product. Your work is the competitive moat.

What We Offer

- Competitive salary or contract rate based on experience and location.
- Fully remote with flexible working hours.
- Direct influence over product direction and AI strategy.
- A technically fascinating problem space — autonomous test generation is one of the most practical and immediate applications of LLMs in developer tooling.
- Equity or performance-based compensation (for full-time hires).
- Path to AI/Engineering Lead as the team scales.

How to Apply

Send the following to [[email protected]]:

- GitHub / Portfolio — links to relevant projects, open-source contributions, or published work.
- AI Project Brief — a short write-up (5–10 sentences) of a production AI/LLM system you built or significantly contributed to. What was the problem, how did you use the model, and what were the measurable outcomes?
- Availability & Rate — your start date, full-time vs. contract preference, and expected compensation.

Shortlisted candidates will be invited to a 60-minute technical deep-dive — no take-home assignments. We'll discuss your AI system design experience, walk through a prompt engineering scenario relevant to our domain, and talk about how you'd approach our current bottlenecks.

We are an equal prospect employer. We welcome applications from candidates of all backgrounds.

Pay: From ₹341,000.00 per year

Work Location: Remote

📌 Senior AI Engineer — AI-Powered Test Automation Platform (India)
🏢 Radixlink
📍 India