Staff Machine Learning Engineer
Salary: USD 154k–216k · Employer-posted
USD154k–193kper year
Rivian’s AI team sits within the Analytics function of the Customer organization — the team that spans Marketing, Sales, and Market Intelligence — and is building the next generation of intelligent, autonomous systems that connect these functions into a single, reasoning ecosystem. As a Staff/Lead Data Scientist focused on Context Engineering, you will own the quality, safety, and economics of what AI agents retrieve and generate — building the evaluation frameworks, observability tooling, security and isolation controls, and memory systems that determine whether an agent’s answer is accurate, trustworthy, and worth what it costs to produce. This is a deeply technical, high-impact role at the frontier of applied AI, working closely with Agentic Solutions, Data Platform, and the broader Customer org Analytics team to bring agentic intelligence to production.
Retrieval Quality & Observability
Retrieval Quality & Evaluation: Build and own the evaluation framework for retrieval and generation quality — benchmarks, golden datasets, human-in-the-loop review, and production feedback signal — and use it to continuously audit the retrieval pipeline, closing the gap between what’s retrieved and what the agent needs.
Context Observability & Production Ownership: Instrument the retrieval-to-answer path end to end so a bad answer can be root-caused to a specific retrieval, chunk, or prompt decision, and carry operational responsibility — triage, rollback, root-cause follow-up — when quality degrades in production.
Data Freshness & Re-Indexing: Design the re-indexing and cache-invalidation strategy so fulfillment and sales data reflect real-time state rather than a stale snapshot — a distinct failure mode from relevance.
Context Economics, Security & Memory
Token Economics & Context Budget: Treat every token as a cost — design context compression, prompt caching, and dynamic context windowing to keep latency and inference cost down without degrading answer quality.
Episodic & Context Memory: Design how agents remember across sessions — what’s retained, summarized, or forgotten — so prior context informs reasoning without overwhelming the context window.
Education: Bachelor’s degree in Computer Science, Information Systems, Data Engineering, or a related field; Master’s preferred.
Experience: 8+ years in Applied ML, NLP, or ML Engineering operating production RAG or LLM-based retrieval systems.
Evaluation & Benchmarking: Experience designing LLM evaluation frameworks — automated and human-in-the-loop — to measure accuracy, groundedness, and hallucination rate, and track drift over time.
Production Operations & Observability: Experience instrumenting and tracing LLM/RAG pipelines (e.g., LangSmith, Arize, Phoenix) to debug production issues, carrying on-call ownership of quality or security incidents, and feeding production signal back into evaluation and freshness strategy.
Context Security: Experience defending LLM systems against prompt injection and data leakage, including isolation, sandboxing, and authorization enforcement at the retrieval layer.
Technical Mastery: Strong command of vector databases and retrieval libraries (e.g., LangChain, LlamaIndex, FAISS) and deep fluency in end-to-end retrieval pipelines. Expert-level Python, strong SQL.
Token Economics: Experience optimizing LLM systems for cost and latency — context compression, prompt caching, dynamic context windowing — at production scale.
Memory Systems: Experience designing episodic or long-term memory for conversational agents — what persists, how it’s compressed, how it’s retrieved.
Environment: Comfortable in a fast-paced, high-ambiguity environment with a builder’s mentality and the ability to make principled decisions with incomplete information.
Salary: USD 154k–216k · Employer-posted
Salary: USD 152k–277k · Employer-posted
Salary: USD 151k–241k · Employer-posted