ADP is hiring a Senior Data Scientist - Ventures - We're seeking an experienced Senior Data Scientist to lead the development of next-generation conversational AI systems. - You'll work to architect end-to-end pipelines — spanning from prompt design through model orchestration and deployment — to deliver cutting-edge text-based assistants at scale. Key Responsibilities - GenAI & LLM Integration: Design, prototype, and productionize solutions using OpenAI, Anthropic Claude, and open-source large language models. - Function & Tool Calls: Implement robust function-calling and tool-calling workflows to connect your LLM agents to external APIs, knowledge bases, and microservices. - Conversation Management: Build and maintain conversation history stores and memory to enable context-aware dialogue. - Prompt Engineering & Testing: Lead iterative prompt design, to build the most optimized custom prompts for different tasks. - Agent Orchestration: Evaluate, select, and integrate agent frameworks (e.g., LangChain, LangGraph, SmolAgents, or similar) to coordinate multi-step workflows and dynamic tool usage. - Model Fine-tuning: Fine-tune transformer models for custom tasks, including text classification, information extraction, and named-entity recognition (NER). - Conversation Engine Development: Build and maintain scalable pipelines for conversational assistants. Must-Have Qualifications - Experience in Machine Learning/AI, focused on large language models and conversational systems. - Hands-on expertise with OpenAI APIs (ChatGPT, function call) and/or Anthropic Claude; familiarity with open-source LLMs (e.g., Llama, Mistral). - Demonstrated ability to build custom prompt templates. - Proficiency with at least one agent orchestration framework (LangChain, LangGraph, SmolAgents, or similar). - Solid understanding of session management, conversation history/data storage, and memory architectures. - Experience fine-tuning transformer models (Hugging Face Transformers or equivalent) for classification, extraction, and NER tasks. - Strong software engineering skills: Python, version control (GIT), test-driven development, CI/CD. Nice-to-Have - Familiarity with Model Context Protocol (MCP) for context management across chains and sessions. - Expertise with Pydantic for response validation and structured responses. - Experience with Retrieval-Augmented Generation (RAG) workflows — including vector stores, embedding search, and retrieval pipelines. - Background in scalable microservices or serverless architectures for deploying AI agents. - O modelo de trabalho adotado pela ADP é office-based/presencial, com a possibilidade de trabalho em home-office por até duas vezes na semana. A little about ADP: We are a comprehensive global provider of cloud-based human capital management (HCM) solutions that unite HR, payroll, talent, time, tax and benefits administration and a leader in business outsourcing services, analytics, and compliance expertise. We believe our people make all the difference in cultivating a down-to-earth culture that embraces our core values, welcomes ideas, encourages innovation, and values belonging. We've received recognition for our work by many esteemed organizations, learn more at ADP Awards and Recognition. Diversity, Equity, Inclusion & Equal Employment Opportunity at ADP: ADP is committed to an inclusive, diverse and equitable workplace, and is further committed to providing equal employment opportunities regardless of any protected characteristic including: race, color, genetic information, creed, national origin, religion, sex, affectional or sexual orientation, gender identity or expression, lawful alien status, ancestry, age, marital status, protected veteran status or disability. Hiring decisions are based upon ADP’s operating needs, and applicant merit including, but not limited to, qualifications, experience, ability, availability, cooperation, and job performance.