octaviokh

As an AI Developer i have worked in the development of LLM’s, Model Training, Tuning & Deployment, Sentiment & Data Analysis using ML and NLP, Multi Model integration (OpenAI, SERPapi, Anthropic, Google, Haystack) and serverless architecture based on AWS. Worked on migrations, architecture, MLOps, workflows and deployments through GitHub actions.

As a data engineer, I built to the development of internal libraries for the AI Sales Assistant using Python as the main language, focusing on scalable systems that supported Delta Tracking, Document Ingestion, ETL, Vector Enhancement, etc. I implemented multiple microservices built on RAG architecture to enhance the data infrastructure and improve updated access to contextual knowledge across the US and Brazil sale departments. Worked on the architectural design and maintenance of the data pipelines.

As a LLM Trainer, i defined interactions between the agent and the user, and created a variety of function pipelines, designed tasks and system prompts, trained the agent with the correct tool usage, corrected responses and redirected tool calls, assigned scores and reviews to my peers.

As a Full-Stack Developer at TV AZTECA, we migrated the internal system of the TV giant. During the Domain Driven Designed project we developed and deployed the Billing and Collections module, a Microservice which used an Event Based pattern with Python, Typescript, Kafka, PostgreSQL and AWS Services, later on i lead a group of developers and delivered the Payment Gateway module, a Serverless/CQRS Microservice based on Hexagonal architecture pattern which reduced average invoice payment time by over 40% percent. I built the architecture and developed a custom model that does a customer segmentation using AWS sagemaker and related services.

As a Full-Stack Developer at Sodexo, we built a web application providing employee benefits for companies. Our project used Typescript for the Front End and Java for the backend on a monolithic architecture. We developed and integrated APIs, implemented biometric validation and a payment system delivering a robust and user-friendly platform. Additionally, we employed tools like Azure and Docker to streamline deployment and ensure a scalable, efficient development environment.

Web administration and maintenance for all the Entretenimiento ECO Group which comprises over 25 sites. Developed over 15+ applications, plus +10 internal services for the group mostly with Python and React, Administer Linux-based webhost, domains and hosting services management, SSL certificates and program updates to prevent downtime. Support maintenance and creation of new server instances, Database maintenance and table creations. Assist users with access requests. Logging/Reports & Debugging. Add / Remove Users (RBCA/ACL)