Gervasio Ramírez

Data Analyst @Peakload

London, GB

Lead the design and implementation of retrieval-augmented generation (RAG) workflows that connect internal PDFs, tables and analytical datasets in Redshift/S3 to LLMs via AWS Bedrock and OpenSearch, exposing them through internal APIs and chat-style tools for querying company knowledge.Build and deploy these systems in Python using LangChain, Hugging Face Transformers and deep learning frameworks (PyTorch, TensorFlow), with a solid understanding of RAG and LLM orchestration techniques.Evaluate and optimise models and retrieval components for accuracy, latency and cost, using tools such as RAGAS and LangSmith and applying an optimisation mindset to prompts, ranking, caching and feedback loops.Collaborate with business and technical stakeholders to map existing workflows and replace them with production AI pipelines that automate several hours of manual analysis, improving speed and consistency.

Universidad de Montevideo

Grado en Ingeniería, Ingeniería industrial