Data Scientist Associate
JPMorgan Chase9 hours ago
Bengaluru, IndiaMid Level
Responsibilities
- Build and productionize end-to-end RAG applications for intelligent search, question answering, summarization, and workflow assistants.
- Implement Agentic RAG patterns including query planning, multi-hop retrieval, tool-based lookups, reranking, guardrails, and fallback behaviors.
- Develop LLM-based NLP capabilities for classification, extraction, summarization, semantic search, and conversational flows.
- Prepare and curate datasets through cleaning, deduplication, normalization, metadata enrichment, and labeling.
- Run applied data science experiments and compare prompts, retrieval methods, embedding models, chunking strategies, and rerankers.
- Define and monitor retrieval and generation quality metrics and create lightweight dashboards and reporting.
- Build analytics pipelines using usage and quality signals such as feedback, clicks, escalations, latency, and cost.
- Create testing and evaluation harnesses with golden question sets, regression tests, adversarial prompts, and safety checks.
- Collaborate with product, design, and stakeholders to translate requirements into shipped features.
- Apply security, privacy, responsible AI, access-control, logging, and audit requirements.
Requirements
- At least 3 years of experience in software engineering, applied machine learning, data science engineering, or a related role building production systems.
- Strong Python programming skills and experience with APIs and services.
- Working knowledge of data cleaning, exploratory data analysis, basic statistics, evaluation design, and communicating results.
- Experience developing RAG applications with frameworks such as LangChain or LlamaIndex, or equivalent experience.
- Comfort with SQL and data tools such as pandas and Spark for dataset preparation and analysis.
- Experience with AWS or Azure and standard software development lifecycle practices, version control, continuous integration and delivery basics, and testing.
- Preferred experience with vector databases or search systems such as OpenSearch, Elastic, Pinecone, Weaviate, or FAISS.
- Preferred experience with offline relevance labeling, LLM-as-judge evaluation with guardrails, regression suites, and basic experiment design.
- Preferred familiarity with agent frameworks such as LangGraph or Semantic Kernel and Agentic RAG workflows.
- Preferred familiarity with embeddings and retrieval concepts.
Tech Stack
Categories
About JPMorgan Chase
JPMorgan Chase provides consumer and commercial banking, payments, credit card, wealth management, and corporate and investment banking services to individuals, businesses, institutions, and governments. The public company (NYSE: JPM) earns revenue from interest, fees, trading, and asset management across operations in more than 100 markets. Headquartered in New York City with roots dating to 1799, it serves retail customers and prominent corporate and government clients through brands including Chase and J.P. Morgan.