AI Research Engineer

AI Research Engineer

Echelon AI · Full-time · In-person - San Francisco

Who we are

About Echelon

Echelon is an AI delivery platform for ServiceNow. We build AI agents that plan, implement, and validate ServiceNow solutions, compressing weeks of consultant effort into hours without sacrificing quality. Fortune 500 companies and the ServiceNow partners who serve them already use Echelon to clear their backlogs: one major CRM provider unlocked 4,000 developer hours and cleared more than 1,000 tickets, a Fortune 10 healthcare company took a CSM migration from eight months down to ten weeks, and a large insurance provider replaced a year-plus migration with a ten-week ITSM go-live, executed correctly the first time.

Echelon was founded by the AI leadership team behind Moveworks, which was acquired by ServiceNow for $3 billion, and is backed by leading investors and executives from ServiceNow, OpenAI, Meta, and Salesforce. We're a ServiceNow Registered Partner for both Consulting & Implementation and Build, and we're SOC 2, ISO 27001, and GDPR compliant.

The role

Part scientist, part systems engineer. We're looking for a research engineer who will research and build long-horizon AI agents that perform end-to-end enterprise change delivery, from planning through testing and deployment, safely and autonomously. You'll build the foundation of autonomous enterprise AI: the reasoning, learning, and safety core that makes digital labor real.

The challenge: foundation models can describe changes, but they can't yet own them. You'll invent new reinforcement learning, planning, and evaluation methods that enable AI agents to handle real enterprise complexity, including multi-step reasoning, compliance guardrails, and measurable outcomes. Your work generalizes across thousands of customer workflows, and your fingerprints will be on how autonomous agents deliver enterprise change, not just at Echelon but across the industry.

What you'll do

• Build RL-driven agent architectures that plan, execute, and validate enterprise platform changes across ServiceNow, SAP, and NetSuite

• Design hierarchical planners that decompose ambiguous business requests into executable workflows

• Develop multi-agent coordination between planner, executor, and verifier roles with shared scratchpads and safety guardrails

• Implement post-training and fine-tuning pipelines across heterogeneous enterprise data such as schemas, ACLs, workflows, and integration patterns

• Build RL and preference optimization loops (offline RL, DPO, PPO, RLAIF) that target business outcomes, not token accuracy

• Create simulation sandboxes and gym environments that safely evaluate agents pre-deployment

• Author eval suites that replicate real customer requests, regression tests, and compliance scenarios

• Build metrics that connect agent decisions to measurable business KPIs and surface attribution automatically

• Engineer reliability via deterministic replays, rollback policies, and isolation boundaries for PII and compliance

• Build model inference and serving pipelines that optimize compute across agent roles

• Partner with applied engineers and product to deploy learning loops to Fortune 500 environments

• Document, publish internally, and set standards for reproducibility, red-teaming, and experiment hygiene

Who you are

We're looking for someone who meets the minimum requirements below. If you meet them, you're encouraged to apply. The preferred qualifications are a bonus, not a requirement.

Minimum requirements

• Research engineer mindset, blending deep ML/RL intuition with strong software engineering fundamentals

• A builder who has shipped ML or LLM systems that work in production, not just papers or prototypes

• A curious experimentalist who instruments, tests, and iterates until it works safely and reliably

• A systems thinker who sees the whole stack: model, API, sandbox, deployment, and metrics

• Outcome-driven, optimizing for business impact rather than benchmark scores

• ML / RL / LLM experience with strong Python skills

• Comfort with typed backends, async pipelines, and integrations

• Excellent communication with both technical and non-technical stakeholders, and comfort operating in an ambiguous, fast-paced environment

Preferred qualifications

• Experience with fine-tuning, tool use, multi-agent orchestration, or enterprise-scale RL

• Startup or fast-paced environment experience

Why Echelon is the right move

• Hard problems, real stakes: Apply RL and reasoning to high-value, regulated enterprise environments.

• Category creation: Define how autonomous agents deliver enterprise change.

• Massive leverage: Each breakthrough generalizes across thousands of customer workflows.

• Founding-team impact: Work directly with the founders and shape Echelon's research roadmap.

• Learn from the best: Direct collaboration with ex-Moveworks AI leaders building autonomous enterprise AI.

Compensation & Logistics

• Competitive, fair-market compensation based on your experience and location

• Eligible for meaningful early-stage equity

• Fully remote, with occasional travel for team gatherings (travel and logistics costs covered by Echelon)

• You'll work daily with frontier AI models and our own agent platform

How we hire (fast and human)

• Intro call with the team (30 minutes)

• Technical deep-dive on your ML/RL experience, plus a working session

• Meet our founders to learn about the vision and the strategic projects ahead

• Decision, typically within two weeks of first contact