QA Engineer, AI Products
The Opportunity Since 2005, MDCalc has been an essential part of the clinician’s workflow to help achieve better patient outcomes. Actively used by more than 65% of physicians worldwide, MDCalc is the most broadly used medical reference – at the point-of-care – for clinical decision tools and content, and one of only four references used by >50% of US HCPs. These evidence-based tools and content are used by millions of medical professionals globally and support 50+ specialties and cover 200+ patient conditions. To continue to further accelerate and steward this growth, we are expanding the AI product team with a QA Engineer. This role will be critical to MDCalc’s expanded success in continuing to support our millions of clinical users worldwide in taking care of hundreds of millions of patients. The Role As a QA Engineer on the AI Products group at MDCalc, you will play a key role in ensuring the quality, reliability, and clinical trustworthiness of MDCalc's AI-powered features. You'll focus on the unique challenges of testing LLM-based systems, where outputs are non-deterministic, correctness is often a spectrum rather than a binary, and regressions can be subtle. You'll be part of a collaborative, fast-moving team that takes pride in delivering software that clinicians trust to care for millions of patients worldwide. The responsibilities of this individual include the following, but are not limited to: Design and execute test strategies for LLM-powered features, including prompt regression testing, output evaluation, and hallucination detection Build and maintain automated evaluation pipelines (eval sets, golden datasets, LLM-as-judge frameworks) to catch quality regressions in non-deterministic outputs Perform black-box and exploratory testing of MDCalc's AI features across web and mobile, with particular attention to clinical accuracy, safety, and edge cases Define quality metrics for AI outputs (accuracy, faithfulness, relevance, safety, latency, cost) and establish thresholds for release readiness Collaborate cross-functionally with engineers, product managers, ML/AI engineers, and clinical reviewers to define what "good" looks like for AI responses Investigate and triage AI failure modes, distinguishing model issues, prompt issues, retrieval issues, and integration bugs Participate in team discussions, offering feedback on testability, risks, prompt design, and guardrails Help develop QA strategies to expand future testing capacity, automation, and evaluation coverage as the AI product surface grows Your Background 5+ years of experience in software QA, with at least 1 year of hands-on testing of LLM-based or AI/ML-powered features Strong understanding of QA principles, test case creation/documentation, and best practices for both deterministic and non-deterministic systems Hands-on experience with LLM tooling and concepts: prompt engineering, RAG systems, evaluation frameworks (e.g., Promptfoo, Braintrust, LangSmith, DeepEval, Ragas, OpenAI Evals), and LLM APIs (OpenAI, Anthropic, etc.) Experience designing automated qualitative evaluation approaches, including LLM-as-judge, rubric-based scoring, semantic similarity checks, and golden dataset regression testing Proficiency with test automation tools, with a focus on Playwright Strong SQL skills for data validation, test data creation, and verifying data integrity across systems Familiarity with token usage, latency profiling, and cost monitoring as quality signals Eagerness to learn quickly and a positive, solutions-oriented attitude Clear and concise communicator, able to surface issues, blockers, and risks effectively when communicating ambiguous or probabilistic failures Self-motivated, proactive, and able to manage time and priorities independently What MDCalc offers: Ability to make a true difference in medicine: MDCalc is the most broadly used medical reference by physicians, used by over 65% of US attending doctors weekly Medical, Dental, & Vision Coverage, with option to extend to your dependents Company-sponsored short-term insurance Fully-paid 8 week parental leave, after 6 months of employment Company-sponsored 401k, after 3 months of employment Unlimited vacation for salaried roles - we trust you to take the time you need Bi-annual company offsites to connect, reflect, and plan together Work from home monthly stipend A culture of fun and motivated team members who believe in a greater mission here at MDCalc Apply To This Job