Staff Machine Learning Engineer

ServiceNow•Santa Clara, CA

About The Position

We build the AI layer of our CPQ platform — a set of Python services that let users configure, quote, and manage transactions through natural language instead of forms. This isn't a thin LLM wrapper. We're running multiple production agent architectures concurrently (ReAct-style tool-calling agents, hand-rolled LangGraph state machines, and the Harness — our from-scratch, industry-leading agent execution runtime). Our systems are backed by a first-party MCP surface into admin/product/rules/transaction systems and interoperate with other AI agents over the A2A protocol. Below that sits a conventional Java/Spring Boot microservices fleet and a React/TypeScript + Lit frontend that the agents ultimately drive. We're looking for someone who already operates at a Senior-Staff bar in the agentic/LLM domain but is building out breadth across the rest of the stack. You'll be one of the most senior technical voices on how agentic systems get designed here — state management, tool boundaries, streaming protocols, prompt/context architecture, and multi-agent coordination — while staying credible end-to-end: able to read a Spring Boot service, unblock a frontend integration, or reason about a classical ML model pipeline when the problem calls for it.

Requirements

8+ years building production software, with several years specifically shipping LLM-powered / agentic systems (not just API wrapper calls — real tool-use loops, state management, multi-turn orchestration)
Experience in leveraging or critically thinking about how to integrate AI into work processes, decision-making, or problem-solving.
Deep, hands-on expertise with LangGraph and/or LangChain (or the judgment to know when to skip them and hand-roll something better)
Strong understanding of MCP — ideally having built an MCP server, not just consumed one
Production async Python (FastAPI, asyncio) — comfortable with WebSockets, streaming, and the failure modes of long-lived stateful connections
Track record of making real architecture decisions on agent systems — tool boundaries, context/state design, cost/latency tradeoffs, when a bounded agent beats a fully autonomous one
Enough range outside Python to read/modify a Spring Boot service and a React component without hand-holding — this is explicitly a whole-stack role, not "Python specialist with a frontend allergy"
Comfort operating with ambiguity and setting technical direction, not just executing a spec — this is a Staff-level bar on judgment
A demonstrated habit of pulling the latest from the industry — new agent frameworks, protocol standards, model capabilities — into production quickly
Willingness and ability to build real fluency in the Core ServiceNow platform.
Willingness to work directly with customers/deployments as part of Forward Deployed Engineering efforts

Nice To Haves

Experience with classical ML (PyTorch/scikit-learn) in addition to LLM-based systems
Experience with A2A or other agent-to-agent interop protocols
Experience with RAG / knowledge-graph systems (embeddings, vector or graph-based retrieval)
Multi-tenant SaaS experience, especially around per-tenant isolation of stateful connections/resources
CPQ, quoting, or transaction/pricing domain experience
Prior Forward Deployed Engineering experience, or time spent embedded with customers shipping bespoke solutions exposure

Responsibilities

Design and implement multi-agent orchestration systems using LangGraph/LangChain agents over frontier LLMs for transaction editing, conversational configuration, and multi-product quote planning.
Develop and own the evolution of 'The Harness', our proprietary agent execution runtime, focusing on full-duplex sessions, low-latency streaming, backpressure, live progress, partial results, and graceful cancellation.
Influence the strategic direction of MCP as a secondary interface versus the Harness as the primary agent interaction mechanism.
Contribute to the A2A protocol for agent-to-agent task delegation and streaming.
Engage in Forward Deployed Engineering, working with customer- and product-facing teams against live deployments to adapt the Harness and agents to real-world workflows and constraints.
Develop and refine RAG / context engineering strategies, including tenant-uploaded document ingestion, categorization, and aggregation into agent context, and prefix-cacheable prompt design for cost/latency.
Evaluate and potentially extend classical ML pipelines (PyTorch/scikit-learn) for problems not suited for LLMs, and compare them against LLM-based alternatives.
Demonstrate full-stack fluency by reading/modifying Spring Boot/Java services and React components to unblock end-to-end integrations.
Operate with ambiguity and set technical direction for agent systems, making architectural decisions on tool boundaries, context/state design, cost/latency tradeoffs, and agent autonomy.
Stay current with industry advancements in agent frameworks, protocol standards, and model capabilities, and integrate them into production systems.
Build fluency in the Core ServiceNow platform to enable direct interoperation with its systems.
Participate in Forward Deployed Engineering efforts, working directly with customers and deployments.