(usa) Senior Manager, Data Science

Walmart · Retail · Bentonville, AR

Senior Manager, Data Science role focused on building and extending an agentic AI platform (Midnight AI) for Walmart's Finance organization. The platform uses autonomous AI agents to research, reason, and respond to requests, integrating with various enterprise data systems and specialized sub-agents. The role involves designing new agents, hardening existing ones, expanding integrations, and ensuring scalability, with a strong emphasis on AI-first development practices and responsible AI.

What you'd actually do

  1. Own and extend a hierarchical multi-agent system built on Pydantic AI and Claude, featuring: Tier 1 — Email Orchestrator: Slim agent handling email I/O, attachment parsing (PDF, DOCX, XLSX with OCR), and response formatting. Tier 2 — Analytical Agent: Heavy-lifting research agent with direct access to BigQuery, Confluence, and web search. Tier 3 — Specialist Sub-Agents: Power BI (DAX generation & dimensional analysis), Code Analyst (sandboxed Python execution & chart generation), and Forecast Agent (Prophet-based time series).
  2. Design and build new sub-agents as the platform expands into new Finance domains.
  3. Evolve our 40+ AI tools spanning BigQuery, Power BI DAX, Confluence, web browsing, data export, geography mapping, and more.
  4. Maintain and improve our LLM-powered learning system — a feedback loop where the agent extracts knowledge from user interactions and applies it to future responses.
  5. Partner directly with Finance Leadership to identify high-impact automation opportunities across Supply Chain, Operations, Merchandising, and Customer.

Skills

Required

  • Python
  • SQL
  • LLM frameworks
  • prompt engineering
  • tool-use patterns
  • agent orchestration
  • BigQuery
  • Power BI
  • enterprise analytics tools
  • finance domain knowledge
  • cross-functional collaboration

Nice to have

  • Pydantic AI
  • Claude
  • OCR
  • Prophet-based time series
  • Code Puppy

What the JD emphasized

  • production-grade, multi-tier agentic system
  • autonomous AI agents
  • without a human in the loop
  • AI-first development
  • Responsible AI is non-negotiable
  • P&L-critical use cases

Other signals

  • production-grade agentic system
  • autonomous AI agents
  • enterprise scale