Full-stack development for Advanced Insights at Atento, building multi-tenant FastAPI APIs, AI pipelines, RAG flows, an analytical data warehouse, and high-performance Next.js web interfaces.
- Developed multi-tenant FastAPI APIs (Python, PostgreSQL, Alembic) for Advanced Insights, exposing feedback services, chat, corpus, monitoring, and dashboard, with JWT authentication and permissions per section/client.
- Implemented the AI Feedback pipeline: automatic feedback generation for support agents with Azure OpenAI, structured outputs (Pydantic), Azure queues, and asynchronous processing from monitoring and transcriptions, with personalized and multilingual action plans.
- Built the chat backend with AI agents and LLMs (VOC), including model and agent registration, Kubernetes (orchestrator + sidecar with tool calling) pod orchestration, and storage of interactions in the corpus.
- Implemented RAG flows over media corpus — embeddings, retrieval, and map-reduce with LLMs — to answer questions about transcriptions and summaries at scale, in a multi-tenant environment.
- Modeled the analytical data warehouse (star schema) and dashboard API: event-driven ETL of Parquet files in Azure Blob to partitioned PostgreSQL, with KPI aggregations (CSAT, TMO, sentiment, volume) and period-over-period variation.
- Developed and optimized responsive, accessible, high-performance web interfaces with React.js and Next.js, prioritizing UX/UI, state management with Zustand, and internationalization.
- Implemented secure authentication and identity integration with Microsoft Authentication (MSAL) in React, ensuring controlled platform access.
- Used Microsoft Clarity to analyze behavior of approximately 6,000 users per continent, optimizing UX of digital journeys and AI channels.
- Refactored React applications by migrating to Zustand, improving performance, scalability, and maintainability.
- Transformed components into server-rendered Next.js pages, increasing performance by around 40% and enhancing SEO.