Overview
AI Evals Engineer role in MSD's Hyderabad Tech Center, designing and running evaluations of AI systems across models, RAG pipelines and agentic applications.
What you'll do
- Design, build, and run evaluations of AI systems across the stack, from foundation and fine-tuned models to retrieval-augmented generation (RAG) pipelines and agentic applications, assessing quality, accuracy, robustness, and safety
- Develop evaluation datasets, test suites, metrics, and scoring rubrics, including automated and LLM-as-a-judge approaches, and validate them against human judgement
- Apply and adapt established evaluation frameworks and benchmarks, and build reusable evaluation pipelines and tooling that integrate into AI engineering delivery
- Analyze and interpret evaluation results, identify failure modes, and communicate findings and recommendations to engineering teams and stakeholders to de-risk AI delivery
- Collaborate with AI engineering, product, and risk and governance teams to define acceptance criteria and evaluation standards for AI use-cases
- Participate in internal professional communities of practice and mentor colleagues to raise AI evaluation and data science maturity within the company
- Proactively scan and keep up to date with AI evaluation research, frameworks, and best practices
What you'll need
- Bachelor's or master's degree in computer science, Artificial Intelligence, Machine Learning, Data Science, Engineering, or a related science or technology discipline
- 3 to 7 years of experience in fields related to AI and ML evaluations, testing and benchmarking
- Qualifications in quantitative, scientific and technology disciplines such as Mathematics, Statistics, Analytics, Physics, and Bioinformatics are appreciated; evaluated on a case-by-case basis
- Proficiency in GenAI, NLP, and ML frameworks
- Experience working on AI and ML projects, with hands-on experience evaluating AI systems such as models, RAG pipelines, or agents
- Awareness of AI evaluation frameworks and approaches, such as benchmark-based evaluation, human evaluation, LLM-as-a-judge, RAG metrics (e.g., faithfulness, answer relevance, context precision and recall), agent task-completion and trajectory evaluation, and safety or red-team testing
- Hands-on code development experience and familiarity with the Python scientific computing stack and evaluation tooling
Nice to have
- Experience using AI testing and evaluation platforms, such as AI red-teaming and security testing tools (e.g., SPLX, Mindgard, HiddenLayer, Lakera, Adversa AI, Enkrypt AI, Pillar Security, General Analysis, Giskard, Protect AI Recon, TrojAI, VirtueRed, NeuralTrust, Repello AI, Garak, PyRIT, Promptfoo), LLM, RAG, and agent evaluation tools (e.g., DeepEval/Confident AI, Ragas, LangSmith, Arize Phoenix, Galileo, Braintrust, LangWatch, Langfuse, TruLens, MLflow, OpenAI Evals, Inspect AI), or AI assurance and governance testing tools (e.g., Resaro, AI Verify, Credo AI, Holistic AI, IBM AIF360)
- Experience in qualitative and quantitative research methods (e.g., study design, statistical analysis, annotation and coding of qualitative data)
- Familiarity with developing and deploying within a cloud-based infrastructure and services (AWS or MS Azure environment)
- Domain knowledge in pharmaceutical business areas such as research, clinical, manufacturing, or commercial
- Ability to communicate and express technical viewpoints to senior stakeholders with succinctness and clarity
Details
- Location: Hyderabad (Hitec City Raidurg), India
- Hybrid working model, flexible role pattern
- No travel required
- Regular employee status; domestic relocation
- No visa sponsorship
Read the full description and apply on the company’s own careers page.