About the role JPMorgan Chase’s Asset & Wealth Management Finance organization is building the next generation of agentic AI solutions that act as “digital workers” for forecasting, analytics, and decision support.
As a Senior Data Science Associate, you will design, deploy, and scale large language model (LLM) agents that turn complex finance questions into trusted, actionable insights.
Job responsibilities
- Build production LLM agents for finance workflows using techniques such as retrieval‑augmented generation (RAG), tool use, and multi‑step reasoning.
- Develop robust data and inference pipelines in Python and SQL; integrate agents with APIs, microservices, and BI applications.
- Implement evaluation frameworks and guardrails: offline and online tests, automatic metrics (factuality, grounding, hallucination rate), human‑in‑the‑loop reviews, red‑team testing, and observability.
- Optimize for scale, latency, and cost across cloud environments; leverage vector databases and embeddings for efficient retrieval.
- Partner with Finance, Product, and Engineering to identify high‑value use cases; translate ambiguous problems into measurable outcomes.
- Apply solid ML engineering and MLOps practices (versioning, CI/CD, model registry, monitoring, incident response).
- Document systems, deliver enablement materials, and upskill partners; contribute to standards for privacy, security, and model risk governance.
Required qualifications, capabilities and skills
- 6+ years in data/ML roles, including 3+ years building and operating production ML applications; hands‑on experience with LLMs.
- Strong Python and SQL.
- Practical knowledge of RAG, prompt engineering, fine‑tuning, function/tool calling, and vector stores.
- Experience with cloud platforms (e.g., AWS, Azure, or GCP) and modern data stacks (e.g., Databricks or Snowflake).
- Familiarity with LLM frameworks and orchestration (e.g., LangChain or LlamaIndex) and REST/GraphQL API design.
- Proficiency in analytics and applied statistics; ability to design experiments and evaluate business impact.
- Excellent communication and stakeholder management; comfort working across Finance, Technology, and Operations.
Preferred qualifications, capabilities and skills
- Experience building multi‑agent systems, autonomous workflows, or task planners.
- Eexperience with PySpark or distributed compute.
- Knowledge of model safety, bias, and privacy techniques; experience with model risk management and governance.
- Exposure to observability tools (logging, tracing, telemetry) and A/B testing.
- Background integrating agents with BI/reporting and workflow tools; familiarity with Tableau or similar is a plus.
- Experience with GPUs/accelerators, containerization, and infrastructure‑as‑code.
What success looks like
- 90 days: deliver a pilot finance agent with RAG and evaluation metrics, integrated with key data sources and APIs.
- 6 months: scale agents across multiple workflows, establish guardrails and monitoring, and demonstrate clear improvements in cycle time, accuracy, or user satisfaction.