Sebastian Gomez

Provectus | San Francisco, California - Remote

ML Internship at provectus, focusing on Generative AI. I am currently working on a project that involves building AI features leveraging agentic workflows for intelligent data parsing and processing.

• Optimizing and benchmarking state-of-the-art Large Language Models (LLMs).
• Evaluating and enhancing ML pipelines, encompassing data preprocessing, model training, and rigorous performance analysis.
• Fine-tuning expert models to use in agentic AI workflows.
• Leveraging cloud-based ML services (AWS/GCP) to deploy GenAI solutions into production environments.
• Staying on top of the rapidly evolving GenAI landscape through continuous research.

JustPaid | San Francisco, California - Remote

Full stack development of internal features in both backend (Django) and frontend. I developed flag-based features for users to have access to more capabilities inside the platform. I enhanced the authentication workflow and added user sessions that store information about last connection and last connected account. I developed a suite for founders to have control over the access to the app information at all levels for each user.

• Built AI features leveraging agentic workflows for intelligent data parsing, processing, and structured database entry
• Designed and implemented an agentic pipeline for contract parsing from raw PDF inputs
• Assesses whether input documents contain sufficient data to generate a valid legal contract
• Uses LlamaIndex to extract relevant text content
• Parses extracted data into Pydantic models for Customer, Contract, and LineItem entities
• Integrated a self-validation loop and agent-based decision system to ensure extraction accuracy and data integrity
• Automatically creates and stores validated records in a PostgreSQL database via Django ORM

Universidad de los Andes | Bogotá, Colombia

Conducted laboratory sessions on machine learning fundamentals, covering optimization techniques, linear and logistic regression models, analytical solutions in Ordinary Least Squares (OLS), hyperparameter fine-tuning methodologies, and neural network architectures

• Managed approximately 70 students per semester. Responsible for laboratory resource creation and studentwork assessment.
• Grade given by students: 4.90/5.00

Roche | Bogotá, Colombia

Analytical and detail-oriented Information Solutions Intern with hands-on experience at Roche Colombia, specializing in healthcare data analysis and clinical process visualization.

• Implemented an automatic reports and graphics generator for statistical information about thyroid ill patients.
• Represented workflows for patient exams (pathology and blood) in hospitals into easy-to-understand block diagrams.