Lucas – Python, LLM, RAG

A digital product for automatic refund processing for over 100K users daily.

• Built the Bill Negotiator feature end to end – an agentic system that plans and runs bill-reduction negotiations with providers via autonomous voice agents, covering agent design, action-plan reasoning, and outcome interpretation;
• Designed voice AI agents (VAPI) that place real calls to providers, navigate phone menus and hold loops, and recover from failures with retry logic and an automated evaluation engine;
• Architected the multi-agent orchestration – supervisor routing, tool-calling agents, and human-in-the-loop gates that let agents pause mid-negotiation to collect user data or approvals and resume safely, with Slack integration for live oversight;
• Built a self-hosted, domain-fine-tuned LLM classifier (PyTorch) for a high-load email classification engine, cutting cost while improving accuracy on the email domain;
• Own LLM evaluation and observability across the stack – custom evaluation engines, Langfuse tracing, and prompt optimization to keep agent behavior measurable and reliable in production;
• Led model serving and reliability – LLM serving with fallbacks, exponential backoff, and health checks; resolved recurring production incidents surfaced through Sentry;
• Own the product end to end, including the supporting infrastructure – async email ingestion and an event-driven notification layer (Event Broker on FastAPI + Celery) that keep the agentic workflows fast and decoupled.

Multiple AI agent and LLM-powered projects spanning document processing, conversational assistants, data copilots, and automated report generation.

• Translated business challenges into AI opportunities and defined actionable implementation roadmaps across fintech, insurance, healthcare, legal tech, and marketplace projects;
• Delivered the company’s first AI-enabled product: an intelligent matching system for a services marketplace;
• Designed and delivered a multi-agent AI system for veterinary healthcare using LLMs, agent frameworks, tool-calling architectures, and Langfuse for observability and tracing, covering history management, context orchestration, and external database access;
• Contributed to 30+ AI/GenAI proposals, defining technical approaches, scope, and feasibility during pre-sales;
• Provided technical oversight for multi-phase AI initiatives across regulated healthcare and compliance domains;
• Mentored engineers transitioning into AI development, providing technical guidance and supporting the adoption of AI engineering practices.

Multiple ML and AI projects across sales, business development, and healthcare domains.

• Acted as a trusted advisor to senior stakeholders, articulating the potential of AI and GenAI in transforming business outcomes and fostering strategic partnerships;
• Led the end-to-end development of AI and GenAI products from conceptualization to deployment;
• Collaborated with cross-functional teams, including business development, to define client engagement strategies, provide architectural guidance, and deliver execution roadmaps;
• Developed technical and commercial proposals by understanding client requirements, designing tailored AI and GenAI solutions, and defining execution plans;
• Educated internal teams and external clients on cutting-edge AI capabilities;
• Drove initiatives to build internal critical mass in emerging AI technologies;
• Mentored and guided junior team members, fostering professional growth and advancing organizational AI expertise.

Worked on projects for a large food services global company and a hospitality company

• Worked in multidisciplinary projects, generating value through data science solutions inside product-oriented delivery;
• Understood business problems and translated them into data science problems;
• Led data science solutions end-to-end, from experiment conception to deployment;
• Provided technical leadership to teams building data-oriented solutions;
• Guided stakeholders and teammates toward reasonable decisions concerning data and data science;
• Engaged stakeholders and proposed new data science projects with high business value;
• Prepared and presented Data and AI commercial proposals.
• Mentored peers;
• Participated in MLOps forums for CI/CD best practices for ML.

Brazil’s first social commerce platform, enabling consumers to access lower prices through group purchasing while connecting merchants and customers through a digital marketplace.

• Delivered data science initiatives aimed at improving customer experience and product insights;
• Identified business problems and designed and evaluated data science solutions through experimentation;
• Built an experimental pipeline for cleaning and classifying user-generated comments within the application;
• Mentored junior and mid-level team members, providing technical guidance and support.

Developed the institution’s first ML-based student retention initiative, building a predictive model to identify students at risk of dropping out at the beginning and end of each semester using academic performance and socioeconomic data.

• Designed and implemented data science and ML projects to support corporate business decisions, translating business problems into DS problems;
• Built data extraction and transformation pipelines feeding downstream analytics and models;
• Delivered fast analytics experiments and visualizations to guide decision-making during projects;
• Ran ML experiments with feature engineering, model validation, and testing to support knowledge discovery.