Senior Data Scientist – Generative AI, Agents

• Develop and version agents’ system prompts, including guardrails, permitted actions, flow control, tool-calling policies, output format, and human escalation.
• Prepare and structure knowledge bases for RAG, including cleaning, sectioning, validity metadata, and organization by categories and access groups.
• Evaluate retrieval quality and fine-tune the RAG strategy.
• Build OCR and information-extraction workflows for PDFs, photos, and handwritten documents, with quality metrics and handling for illegible documents.
• Implement anonymization and personal-data masking services, along with input and output guardrails.
• Build golden sets and a prompt evaluation harness, with regression testing for each version and measurement of accuracy and cost per case.
• Evaluate and approve LLMs from different providers, applying AI FinOps practices.
• Create agents on the client’s platform with the technical team and support integration testing, UAT, and assisted operations.
• Document technical decisions and support materials for the architecture committee, covering costs, security, and data handling.
• Work on a project to build five AI agents to automate back-office processes for a healthcare and pharmaceutical retail client.
• Integrate agents with enterprise systems through APIs, MCPs, and RPA automations, with interaction via Microsoft Teams.
• Work in a squad with a solutions architect, developer, and business analyst.

• Bachelor’s degree in Computer Science, Statistics, Mathematics, Engineering, Information Systems, or a related field.
• 4+ years of experience in data science or machine learning, including experience on Generative AI projects.
• Hands-on experience with LLMs: prompt engineering, function calling/tools, structured output (JSON Schema), and agents.
• Experience with RAG: embeddings, vector databases, hybrid search, reranking, and retrieval evaluation.
• Python for experimentation, evaluation, and data processing.
• Building datasets and evaluation metrics for LLM solutions (accuracy, precision/recall, LLM-as-a-judge).
• Knowledge of OCR and document information extraction (multimodal models and/or Document AI).
• Knowledge of AI governance and Brazil’s LGPD: masking, anonymization, and data minimization.
• Intermediate to advanced English proficiency.
• Preferred: Azure OpenAI and/or Microsoft Foundry; MongoDB Atlas Vector Search.
• Preferred: MCP protocol and agent patterns (ReAct, stateful workflows).
• Preferred: LLM security: prompt injection, red teaming, and guardrails.
• Preferred: AI FinOps: estimating and optimizing token costs and comparing providers.
• Preferred: Node.js/TypeScript.
• Preferred: Experience with RPA or integrations with ServiceNow, Microsoft Teams/Graph, or SAP.
• Preferred: Business knowledge in healthcare/regulatory affairs (ANVISA), supply chain, or HR benefits.
• Preferred: Cloud or AI certifications (e.g., Azure AI Engineer).

• Ongoing professional development
• Potential project extension or permanent employment after the project
• Remote work