Software Engineer Iii– AI Systems

Walmart Walmart · Retail · Bentonville, AR +2

Software Engineer III to design and build AI-first systems focusing on agentic AI, high-performance data/compute frameworks, and scalable production services. Responsibilities include building agentic AI services with planning, tool use, and feedback loops, implementing orchestration and guardrails, and collaborating with DS/MLE partners. The role also involves developing GPU-accelerated pipelines with RAPIDS, using Ray for distributed compute, and designing reliable microservices for training/inference, vector indexing, and decisioning. Emphasis on quality, security, MLOps integration, and collaboration.

What you'd actually do

  1. Build agentic AI services (planning, tool use, retrieval, feedback loops) and integrate them with internal systems and APIs.
  2. Implement orchestration, memory, tooling, evaluation, and guardrails for agentic workflows.
  3. Develop GPU‑accelerated pipelines using RAPIDS (cuDF/cuML/cuGraph) and optimize end‑to‑end performance.
  4. Use Ray (or similar) for distributed compute, batch/stream processing, and scalable workflow orchestration.
  5. Design and maintain reliable microservices for training/inference, vector indexing, and real-time decisioning.

Skills

Required

  • Python
  • Go/Java/C++
  • Ray/Spark/Dask
  • RAPIDS (cuDF/cuML/cuGraph)
  • FastAPI/Flask
  • Kubernetes
  • Docker
  • data structures/algorithms
  • concurrency
  • networking
  • systems design

Nice to have

  • agent frameworks (LangGraph-style planners, tool-use patterns, retrieval and memory components)
  • vector databases (FAISS, Milvus, pgvector, Pinecone)
  • feature stores
  • LLM and embedding services
  • Kubernetes autoscaling (HPA/KEDA)
  • GPU scheduling/operators
  • PyTorch profiler
  • Nsight
  • line-profiler
  • Ray dashboard
  • vLLM
  • Triton Inference Server
  • ONNX Runtime
  • TensorRT

What the JD emphasized

  • agentic AI services
  • production-grade services
  • GPU-accelerated pipelines
  • RAPIDS
  • Ray
  • microservices
  • agent frameworks
  • vector databases
  • LLM and embedding services
  • high-throughput inference

Other signals

  • agentic AI services
  • production-grade services
  • GPU-accelerated pipelines
  • distributed compute
  • microservices