> Markdown version of [/jobs/ext/4766353-senior-data-scientist](https://www.wearedevelopers.com/jobs/ext/4766353-senior-data-scientist). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Senior Data Scientist - **Company:** Arrive Logistics - **Location:** Chicago, IL, United States - **Experience:** Expert - **Salary:** $163,000 - $195,000 - **Contract:** Permanent contract - **Skills:** Artificial Intelligence, Code Review, Continuous Integration, Information Extraction, Python (Programming Language), Machine Learning, SQL Databases, Unstructured Data, Data Logging, Ontology Design, Retrieval-Augmented Generation, Large Language Models, Prompt Engineering, Generative AI, Agentic-AI, Information Technology, Production Code, Performance Monitor, Machine Learning Operations, Document Classification, Software Version Control - **Published:** November 18, 2026 - **Apply:** https://www.careerbuilder.com/job-details/senior-data-scientist-chicago-il--dac3b22e-55e1-4255-94bf-75e9fd9b9c57 ## About the Role Data Science, Career Development, Team Player, Transportation and Logistics, Data Science, Customer Relations, Artificial Intelligence (AI) Agents, Taxonomies, Production Systems, Pricing, * Bachelor's or Master's degree in a quantitative field (computer science, statistics, linguistics, or related) and 5+ years of applied ML or data science experience; or a PhD in a relevant field (such as NLP, machine learning, or computer science) and 3+ years of industry experience shipping production ML systems; or equivalent practical experience. * A track record of shipping NLP or LLM-based systems to production and owning them after launch, including evaluation, monitoring, and iteration. * Deep experience with text classification and information extraction, including multi-label problems, taxonomy design, and how these systems fail in practice. * Hands-on experience with prompt engineering, fine-tuning, and retrieval or RAG architectures, with the judgment to choose between them and explain why. * Experience designing evaluation frameworks from scratch: golden datasets, annotation guidelines, label quality processes, and statistically sound offline and online metrics. * Experience building or designing LLM agents or multi-step LLM workflows that take actions in production systems, including guardrails, human-in-the-loop controls, and observability. * Experience setting technical direction for a system or product area, including leading design reviews and making architecture decisions with Engineering partners. * Ability to turn an ambiguous business problem into a scoped roadmap, and to change course when the data says to. * Proficiency in Python and SQL, and comfort working with large volumes of structured and unstructured data. * Solid software engineering fundamentals, including version control, testing, code review, and CI/CD, and the ability to write production-quality code alongside engineers. * Strong written and verbal communication, including presenting trade-offs and recommendations to non-technical audiences. ## Description * Lead discovery into how carrier and customer-facing reps work. Analyze email threads at scale, map them to workflows and decisions, and size where automation creates the most value. * Own the production intent taxonomy across carrier and business development workflows, including how it's validated with stakeholders and how it evolves as the business changes. * Design and develop a real-time, multi-intent classification service that serves multiple downstream agents and products, and partner with Engineering to deploy and serve it with clear accuracy and reliability targets. * Define the evaluation strategy for the platform and the agents: golden datasets, annotation guidelines, offline and online metrics, and the monitoring that tells us when performance drifts. * Work with stakeholders and partners to prioritize which agents to build and in what order, based on volume, value, and feasibility, and define what success looks like for each before it's built. * Design AI agents that act on rep workflows, such as load offering, rate negotiation, capacity management, and shipment execution, and partner with Engineering to ship them and integrate them with Arrive's matching, pricing, and load-tracking systems. * Build guardrails, human-in-the-loop review, and observability into every agent, including tracing, logging, cost tracking, and performance monitoring, so agents act safely in production. * Choose the right approach for each problem, from classical NLP and fine-tuned models to prompting, retrieval, and agent orchestration, and make the trade-offs explicit. * Lead technical design with Engineering and Product, representing Data Science in decisions about where classification and decisioning live. * Consolidate existing email classification systems onto the shared platform over time, reducing duplicated effort across teams. * Mentor other data scientists through design reviews, code reviews, and pairing, and raise the team's standards for building and evaluating LLM systems. * Communicate progress, results, and trade-offs clearly to business and executive stakeholders.