Staff AI/ML Engineer

About Nimbus Data Systems Nimbus Data Systems started with an ambitious mission: to turn the complex process of offering delivery into a simple, turnkey solution. It’s a big mission, and now we want you to join us in making it even bigger. 🚀 We’re proud to be recognized as one of Fast Company’s Best Workplaces for Innovators and a 2025 Inc. Magazine Power Partner , awards that highlight how we’re redefining the future of logistics while empowering our partners to grow. Backed by leading Silicon Valley investors like Village Global, the fund whose investors include Bill Gates, Jeff Bezos, Mark Zuckerberg, Reid Hoffman, and Sara Blakely, we’ve built a world-class team across the globe. We operate at scale but remain small enough for every person to have a massive impact. There’s a lot of important work ahead, and joining Nimbus Data Systems means the opportunity to grow faster than ever while doing the most meaningful work of your career. Here’s a quick overview of what you will be doing: The Role We're hiring a Staff AI/ML Engineer to help build the AI at the core of Nimbus Data Systems — and we mean core, not cosmetic. We're pushing to be at the forefront of AI in logistics: building intelligent agents that automate work the industry still does by hand, and the prediction and optimization models that power smarter operational decisions in real time. This is a high-ownership, hands-on role that spans two connected layers: The agentic layer — multi-agent systems and copilots that understand a request and execute the right workflow end to end, automating manual operational work. The intelligence layer — forecasting, prediction, and optimization models that make and improve the decisions behind the product, getting smarter with every outcome. You'll take ambiguous AI ideas from prototype to production, set the bar for what reliable AI looks like in an operationally demanding domain, and help define what the AI-native version of Nimbus Data Systems becomes. You'll work closely with the founders, product, and engineering, with a high impact on the most important technical work in the company. What You'll Do Design and ship production AI systems — multi-agent orchestration, routing, and specialized agents that take a request and carry it through to a reliable outcome. Automate manual operational work across onboarding, support, exceptions, and document/data understanding — turning processes that take hours or days into seconds. Build the models behind the decisions — forecasting, prediction, matching/allocation, optimization, and reliability scoring that ground the product in data instead of guesswork, exposed as services the agent layer can call. Design the learning loop. Instrument decisions and their outcomes so models continuously improve, with the data and evaluation infrastructure to support it. Own reliability and evaluation. Build the eval harnesses, tracing, observability, and guardrails for complex AI workflows where mistakes carry real operational and financial consequences — and prove a model or agent beats the status quo before it ships. Make the build-vs-rules calls — know when a model genuinely wins, when an agent is the right tool, and when a simple rule is the smarter answer. Raise the bar and help the team grow — push our prototyping-to-production pipeline forward and mentor engineers as the AI team scales. Requirements We care more about evidence of work than pedigree or years of experience — but most strong candidates will have shipped production AI/ML systems at scale. You've shipped production AI/ML, not just prototypes — and dealt with the real tradeoffs of edge cases, quality, latency, cost, and reliability. You have real depth on at least one of these, and working fluency across both: Generative / agentic AI — multi-agent orchestration, tool/function calling, RAG, structured outputs, and the modern stack (e.g., LangGraph/LangChain, MCP), across providers (Amazon Bedrock, Azure OpenAI, Anthropic, OpenAI). Applied ML / decision intelligence — forecasting, optimization, matching/allocation, ranking, or prediction models that drive operational decisions with measurable business impact. You design and trust your own evaluation — offline and online, tied to business outcomes, with safe rollout (e.g., shadow mode) and drift monitoring. You're deeply hands-on and ship fast — strong in Python, modern API/services (e.g., FastAPI), and sound ML-systems and architecture instincts. You've built for operationally complex or high-stakes environments where quality and reliability genuinely matter. You communicate clearly, make decisions quickly, and can lead technical work without needing heavy process. Bonus points: Background in logistics, supply chain, transportation, marketplaces, mobility, or fulfillment. Operations research / optimization, or reinforcement learning / bandits for sequential decision-making. Multimodal / document understanding, computer-use, or browser automation. Real-time / streaming systems, feature stores, and production MLOps at scale. Patents or peer-