Oliver – AI agent development, Python, LLM

An American multinational technology corporation that specializes in internet-related services, artificial intelligence, cloud computing, and consumer electronics.

• Led a cross-functional ML team of 5 engineers and data scientists to develop and productionize an NLP pipeline using Hugging Face Transformers and PyTorch, delivering a 20% improvement in content classification accuracy.
• Designed and deployed a Retrieval-Augmented Generation (RAG) system for video search, integrating LangChain with Gemini 2.0 APIs and VectorSearch.
• Built and monitored production-grade BERT-based models for text classification and recommendation systems, using MLFlow, Argo CD, and Kubernetes for scalable deployment and lifecycle management.
• Researched and implemented multi-modal learning techniques combining text, image, and video embeddings for user behavior prediction, using Tensorflow, and evaluated using LLM metrics.

A global digital engineering company that builds software products and modernizes legacy platforms for tech-driven companies. The firm utilizes advanced data analytics, cloud infrastructure, and proprietary artificial intelligence frameworks to accelerate digital transformation across international markets.

• Integrated Airflow with Spark for extracting SFTP data (time process reduction from 8 hours to 5 minutes).
• Populated production tables in Qubole.
• Generated a risk model (23% improvement in discarding potentially risky clients).
• Created dashboards in Looker for stakeholders' use.

A leading public enterprise software company that provides specialized, cloud-based software and generative AI solutions purpose-built for highly regulated, partner-led industries. Founded in 2000 and headquartered in Palo Alto, California, it designs its technology specifically for professional and financial services firms - including legal, accounting, investment banking, consulting, and private capital firms.

• Managed and optimized production Kubernetes on Azure Kubernetes Service (AKS), ensuring high availability, scalability, security, and fault tolerance.
• Containerized machine learning and data processing applications using Docker, enabling consistent builds and seamless deployments to AKS. Integrated and orchestrated data pipelines using Azure Data Factory, enabling efficient ingestion, transformation, and movement of data to support ML workflows.
• Provided unit testing and integration testing for new deployments, containerized in Docker containers and deployed to production in a Kubernetes cluster (AKS + Azure blob storage + Azure DevOps).
• Created and deployed a Retrieval-Augmented Generation (RAG) system for contextual document search, integrating LangChain with OpenAI/Open source LLMs APIs and custom vector databases on AWS and Azure.
• Designed, enhanced, and maintained GitOps-based CI/CD pipelines using Jenkins, Argo CD, and Helm to automate reliable deployments across multiple environments.

A for-profit social enterprise and financial technology company that builds mobile business tools for micro-entrepreneurs and "mom-and-pop" shops (tiendas) in emerging markets

• Provided data cleaning/analysis and Statistical/ML methods development.
• Maintained a balance between business vision and data/stats.
• Carried out algorithm cleansing and discussions to enhance computational performance and accuracy in predictions.
• Achieved the reduction of cloud calculation time by 94%.
• Worked on the prediction of retail stores' schedules to avoid errors in other algorithms that depend on that data.
• Developed Data Visualization tools that would be commercialized.