Machine Learning Engineer, CX Intelligence

📍 Brazil

BRL 347,900 per year

MIDDLE

✅ Remote

Tech Stack
Tag name is followed by "@" symbol and proficiency level value. About proficiency levels:

1-2 — basic awareness. Minimal hands-on experience, and a rudimentary understanding of the technology's purpose;

3-6 — daily use. Comfortable and regular usage, capable of handling common tasks and challenges related to the technology;

7-9 — you are an expert, you can teach others, you know all the pitfalls and tricks;

10 — exceptional knowledge, comprehensive understanding, and adeptness in all aspects of the technology, including advanced problem-solving. Think twice before claiming or demanding such level.

AI @ 3 API @ 3 Agentic Systems @ 3 Asynchronous Programming @ 6 Customer Support @ 3 GenAI Generative AI @ 3 JSON @ 6 LLM @ 6 Machine Learning Microservices @ 6 Prompt Engineering Python @ 6 R @ 1 RAG React @ 5 Vector Databases @ 6

Details

Coinbase's CX Intelligence Engineering team, part of the Enterprise Applications and Architecture organization, builds the multi-agent platform powering Coinbase Chat, Help Center, and agent tooling. The role focuses on designing and scaling agentic systems that automate complex customer support workflows by connecting large language models with internal APIs and tools to deliver fast, accurate, and compliant AI-powered experiences for millions of customers.

Coinbase is a remote-first, but not remote-only, company. Employees are expected to meet quarterly for in-person working sessions called “surges.”

Responsibilities

Architect multi-agent systems using orchestration frameworks such as LangGraph and Google ADK to automate complex customer support procedures end to end.
Build and scale Model Context Protocol (MCP) integrations connecting LLMs with internal Coinbase APIs, databases, and third-party tooling.
Develop automated “LLM-as-a-judge” evaluation pipelines to monitor, measure, and improve non-deterministic AI agent performance in production.
Implement retrieval-augmented generation (RAG), fine-tuning, and prompt engineering techniques to ensure chatbot responses are grounded, accurate, and compliant with Coinbase policies.
Ship production-ready Python services that are resilient, low-latency, and capable of handling Coinbase-scale traffic across asynchronous microservices.
Partner with Conversation Design and Product teams to translate complex business logic into executable agent procedures within a decentralized architecture.

Requirements

At least 3 years of experience building and scaling production ML or AI systems, including agentic “Loop” or “ReAct”-based systems where AI takes actions autonomously.
Deep understanding of the LLM lifecycle, including context-window management, token optimization, structured output parsing, Pydantic, JSON mode, and multi-agent orchestration frameworks such as LangGraph, Google ADK, or similar technologies.
Strong Python proficiency and hands-on experience with microservices architecture, asynchronous programming, high-throughput APIs, and vector databases such as Pinecone, Weaviate, or equivalent technologies.
Ability to explain model behavior and technical trade-offs to non-technical stakeholders across Customer Experience, Legal, and Product teams.
Ability to use generative AI responsibly, maintain human oversight, and deliver business-ready outputs that improve workflow efficiency, cost, and quality.

Compensation

The annual base salary is R$347,900 BRL, excluding equity and bonus. Total compensation may also include equity, bonus eligibility, and medical, dental, and vision benefits.

Job ID: P77545