About this role
Kensho, S&P Global’s AI innovation hub, seeks a Senior Machine Learning Engineer to develop production-grade ML systems, a generative AI platform, LLM-powered applications, and AI toolkits including Kensho Link and NERD. The role spans retrieval-driven AI agents that ground LLM responses in trusted S&P Global data and broader ML systems designed for reliability, scalability, accuracy, and transparency. The position is listed in Hyderabad and Bengaluru, India.
Responsibilities
- Design, refine, deploy, and operate ML systems that solve business problems and support Kensho products. Build agents that fetch, validate, and structure information from S&P Global datasets to ground LLM answers.
- Evaluate LLM-based agents online and offline; address performance, latency, memory use, compute efficiency, and feature consistency. Scale applications for demand and efficient resource use.
- Work with proprietary structured and unstructured data and subject-matter experts. Own work across problem framing, data exploration, modeling, deployment, monitoring, and continuous improvement.
- Identify and reduce technical debt; scope and execute ML initiatives; contribute to architectural decisions. Collaborate with ML, backend, data, product, design, and engineering teams to improve user workflows.
Requirements
- Bachelor’s degree or higher in Computer Science, Engineering, or a related field, plus 5+ years of significant hands-on industry experience in machine learning, NLP, and information retrieval, including designing, shipping, and maintaining production systems.
- Strong Python proficiency; experience reading SQL databases and writing queries for specific access patterns. Experience building ML pipelines for data processing, training, inference, maintenance, evaluation, versioning, and experimentation, and using LLM orchestration frameworks such as LangChain.
- Effective coding, documentation, collaboration, communication, and problem-solving skills, with adaptability in a fast-paced environment. Experience with RAG-based systems is preferred.
Technologies listed by the team include scikit-learn, XGBoost, LightGBM, PyTorch, Transformers, Hugging Face, Docker, Amazon EKS, Jenkins, AWS, Pandas, Matplotlib, Jupyter, Weights & Biases, DVC, MosaicML, NVIDIA NeMo, Labelbox, PostgreSQL, OpenSearch, SQLite, and Amazon S3. Listed benefits include health coverage, time off, learning resources, retirement planning, a continuing education program, a company-matched student loan contribution, family benefits, and discounts; availability varies by country.