Data Scientist

This is a hands-on role where you'll move from prototype to production quickly and see your work used in the field every day.\n\nWhat you'll be doing\nBuilding and deploying LLM-powered agents and workflows that automate real operational tasks across customer service, dispatch, and back-office functions.\nDesigning and implementing RAG pipelines, tool use, function calling, and multi-agent orchestration.\nIntegrating AI capabilities into the product, with deep connections to Salesforce, internal systems, and third-party APIs.\nPrototyping new AI features quickly, evaluating them rigorously, and shipping the ones that work.\nBuilding evaluation frameworks, prompt versioning, and observability so AI systems stay reliable in production.\nWorking closely with the founder and engineering team to identify the highest-impact problems AI can solve.\nStaying close to the frontier, bringing in new models, techniques, and tools as the space evolves.\n\nWhat we're looking for\nHands-on experience building with LLMs, whether through OpenAI, Anthropic, open-source models, or all of the above.\nStrong software engineering fundamentals, with proficiency in Python and at least one of TypeScript, Node.js, or similar.\nExperience with RAG, vector databases, embeddings, and prompt engineering in production settings.\nComfort with agent frameworks and tool use patterns (LangChain, LlamaIndex, custom orchestration, MCP, or equivalent).\nGood understanding of evaluation, guardrails, latency, cost, and the operational realities of running AI in production.\nComfort working in a small, fast-moving startup, taking ownership from idea to deployment.\n\nNice to have\nExperience fine-tuning models, training embeddings, or working with open-source LLMs.\nFamiliarity with Salesforce, CRM systems, or operations-heavy business software.\nExperience with voice AI, speech-to-text, or telephony integrations.\nBackground in workflow automation, RPA, or no-code/low-code platforms.\nPrevious work on tools for trades, field services, logistics, or other operations-heavy environments.\nOpen-source contributions or side projects in the AI space we can look at.\n\nWhat you'll get\nReal ownership over the AI layer of a product used daily by tradespeople and operations teams.\nThe chance to shape AI architecture, tooling, and best practices at an early-stage startup.\nClose collaboration with founders and a small, senior team, with direct impact on roadmap and priorities.\nA role that combines frontier AI, real-world operations, and meaningful integrations to solve practical problems at scale.\nBudget for compute, tools, and the best models available.\n\nIf you enjoy building with AI, like shipping quickly, and want to use frontier tooling to solve real problems for real users, Chumley's AI Engineer role could be a great fit.