Software Development Engineer in Test

ComPsychⓇ is the worldwide leader in organizational mental health, well-being, and absence management, dedicated to igniting human potential in workplaces across the globe. For over 40 years, we have combined the best in technology with unmatched human expertise to help individuals and their organizations thrive. Our GuidanceResourcesⓇ and AbsenceResourcesⓇ solutions deliver end-to-end mental health, well-being, work-life, health navigation, and absence support to more than 75,000 customers worldwide,  touching more than 160 million lives across 200 countries. Visit compsych.com to find out why 40% of the Fortune 500 choose ComPsych for their mental health and absence management needs.

We're modernizing how ComPsych builds technology — new customer-facing web and mobile experiences, a unified design system, and AI-powered capabilities across our products. That means raising the engineering bar for how we design, build, and operate software, with Engineering, Product, and Design working together end to end.

AI isn't a productivity add-on here — it's how we build. Our team uses agentic development as standard equipment, moving from problem to production-ready code faster than traditional development allows, while keeping the work reviewable and maintainable. We're looking for a Senior SDET who already works this way. Quality here is an engineering discipline, not a phase, and there is no manual QA layer beneath our SDETs: you own the strategy, automation, and signal quality for your team, with exploratory testing as deliberate work alongside maintainable automation. Agents help turn requirements into scenarios, generate automation, and diagnose failures; you own the judgment on whether the evidence is trustworthy. As a healthcare company, PHI never enters non-production environments. Platform owns environments and pipelines, and you own what runs on them: test configuration, data, and signal quality.

This role is for an SDET who has verified AI-powered products whose outputs are non-deterministic. You'll define how ComPsych decides that an AI feature is good enough to ship, and how we know when it stops being so.

AI changes what your time goes toward. You're still accountable for the quality of what ships — strategy, automation, feedback, and data — but your hands-on role in each shifts:

• Define how we verify AI-powered features. Set evaluation methods, thresholds, and release criteria for products whose outputs vary from run to run, and distinguish a real regression from acceptable variation.
• Own the team's quality strategy. Define the verification strategy for a system based on risk, decide what evidence each layer needs, direct exploratory effort at unfamiliar failure modes, and state what is deliberately not tested and why.
• Build the automation and evaluation tooling the team runs on. Design eval harnesses, fixtures, and shared patterns for reliable, repeatable execution across model, prompt, and retrieval changes, accounting for concurrency, asynchronous behavior, and eventual consistency.
• Make test feedback actionable. Own execution speed, gating, failure classification, and flake discipline so developers can self-resolve failures, including telling a flaky test from genuine output variance.
• Own synthetic data and PHI guardrails. Define the synthetic-data strategy, isolation patterns, and CI protections that keep PHI out of non-production, including the evaluation data AI features need, while giving engineers deterministic test state.
• Set testability and non-functional quality where risk warrants. Define environment, observability, and data needs with Platform rather than building a parallel one, and set the depth of safety, security, performance, and resilience verification, including hallucination, unsafe output, and sensitive-data leakage.
• Direct AI-first quality practice, and keep judgment human. Author agent conventions as versioned, reviewable artifacts, set boundaries for what agents may do without review, and treat a generated test that passed once as a draft, not a test.

Everyone in this role brings the foundations below, plus the role-specific depth that follows.

• You're strong in a modern programming language and can reason about concurrency, asynchronous workflows, and failure propagation across services.
• You can trace a failing test across services using logs, traces, and deployment context, and tell whether the root cause is product, automation, data, or environment.
• You communicate residual risk clearly to Engineering and Product, including when shipping with a known gap is the right call, without becoming the release bottleneck.
• You have real depth in at least one non-functional area — performance, resilience, security, or accessibility — and know when to bring in specialists.
• You already work AI-first. You've set conventions other engineers actually followed and can explain where you draw the line on what an agent may change without human review.

• You've verified products built on LLMs or other non-deterministic systems, where the same input can produce different valid outputs.
• You've designed evaluation approaches such as curated eval sets, rubrics, statistical pass thresholds, and human- or model-assisted grading, and can explain how you decide a result is trustworthy.
• You've caught regressions across model, prompt, or retrieval changes, and tested for failure modes like hallucination, unsafe output, and sensitive-data leakage.

Bonus: experience building shared testing frameworks and standards that multiple teams adopted; synthetic-data strategy and deterministic isolation for sensitive data; failure injection or recovery testing in production-like environments; or working in healthcare, another regulated domain, or systems that handle sensitive customer data.

• Full benefits package, including Paid Time Off (PTO), medical, dental, vision, 401(k) with match, robust EAP, wellness program, and much more
• The salary range for this position is $135,000 - $165,000 (USD). The base salary range represents the anticipated low end and high end of the range for this position. The actual compensation will be influenced by a wide range of factors including, but not limited to previous experience, education, pay market/geography, and scheduled hours. 

EEO

ComPsych is an equal opportunity employer.  All applicants will be considered for employment regardless of race, color, age, genetics, religion, gender, sexual orientation, gender identity, national origin, disability or protected veteran status and any other characteristic protected by federal, state or local laws. ComPsych Corporation maintains a drug free workplace.

ComPsych Corporation and its affiliates is committed to the responsible and transparent handling of your personal data. Information collected during the recruitment process will be used to assess your application and, where applicable, to prepare and administer an offer of employment. Your personal data will be processed in accordance with the privacy laws applicable in your country or jurisdiction of residence, which may include the California Consumer Privacy Act (CCPA), the EU General Data Protection Regulation (GDPR), Canada's Personal Information Protection and Electronic Documents Act (PIPEDA) and applicable provincial privacy legislation, and other applicable national or local privacy laws. For full details on what personal data we collect, how we use it, your rights as an Applicant, and how to contact us with privacy-related questions, please review the relevant Applicant Data Privacy Notices: US Applicant Data Privacy Notice & Canada Data Privacy Notice.