Contributing as a core development team engineer for Taco-IDE project, an AI-powered learning platform.

https://drone-comp.ita.br/author/rodrigo-d.-anderson/

At Spectre AI, a Web3 AI startup, I work in a AI Engineer team that is focused focus on building a multi-agent system that executes tasks related to crypto sentiment analysis, technical analysis, and price prediction. The system is integrated with a chatbot on Telegram and the platform's website, providing users with insights and technical explanations powered by AI. I also implemented a RAG (Retrieval-Augmented Generation) architecture to enhance the quality and relevance of the generated responses.

Key Technologies:

GCP, LangChain, LangGraph, vector database (qdrant, FAISS, pgvector, elasticsearch), Vertex AI, FastAPI, Docker, Docker Compose, Kubernetes, Python

At Invillia, an outsourcing company, I work as an AI Engineer on the platform team. My responsibilities include building AI agents using LangChain, vector databases, and the OpenAI API. I developed an AI agent for company managers to interact through a Teams AI chatbot. Currently, I am working on a voice AI agent for engaging with job candidates. This agent uses generative AI and open-source models like LLaMA 3 and will be deployed on Kubernetes in the Azure cloud. My work involves using technologies like Docker, Azure WebApps, Azure OpenAI, CI/CD through GitHub Actions, and Python.

Maintenance and monitoring of machine learning models, alongside the development and refinement of CI/CD pipelines tailored for these models. This effort supports our global marketing team's B2B and B2C strategies, through both API development and batch pipelines, ensuring scalable and seamless deployment. Utilizing Python for algorithmic development, I leverage MLFlow for experiment tracking and model versioning, promoting transparency and reproducibility. Kubernetes is pivotal in sustaining a robust, scalable infrastructure, facilitating efficient model training and deployment processes. Our data management relies on PostgreSQL, supported by Dell Cloud's S3 for extensive dataset storage. The DevOps strategy is critically underpinned by Airflow for workflow scheduling and monitoring, and GitLab for version control and CI/CD enhancements. In this role, I bridge data science, operations, and software engineering, contributing to a diverse and international team spanning Slovakia, the USA, Italy, and India. This collaboration fosters cross-cultural innovation and strategic, data-driven solutions, underscoring our commitment to building and maintaining reliable machine learning models and deployment pipelines for Dell's global marketing initiatives.

Coordinated the ML product team at EmCasa, a proptech startup, to build intelligent systems based on AI. Responsible for the continuous development and deployment of machine learning models in the cloud using AWS, I managed the versioning of models, metrics, and data with tools like EKS, Airflow, and MLflow. My duties extended to constructing REST APIs to serve these models on the site and creating batch pipelines via Airflow for efficient data processing. Additionally, I oversaw code reviews and ensured alignment with the business team's expectations. A key part of my role involved deploying our solutions on AWS, leveraging a suite of services including S3 for data storage, Athena and Redshift for data querying and warehousing, Glue for data integration, and Docker and Kubernetes (EKS) for container orchestration. I also managed our Docker images in ECR and utilized EC2 for computing resources, ensuring our ML infrastructure was scalable, reliable, and aligned with our product and business needs.

Ensure continuous delivery and continuous development of machine learning models in the cloud (Azure). Versioning of: models, metrics and data (Airflow and MLflow). REST API construction, Code review, pull request approval. Alignment of expectations with the business team. Ensure the development of solutions in the data platform with data mesh architecture: Data Infra as Platform.

-Data science consultant at Saint-Gobain (industrial sector) responsible for machine learning projects in the area of digital intelligence Industry 4.0.

Projects: Credit score, factory accident forecast and demand forecast.

- Data scientist in the product: AIJUS - AI applied to the law. Worked as Data Scientist

developing models of neural networks in natural language processing and

text mining.

- Data scientist in the product - Smarter sales: AI applied to retail,

creating models to optimize inventory and reduce losses.

models: demand forecast, Leadtime and rupture forecast by each SKU in Rede OBA - Hortifruti.

Development, maintenance and improvement of the virtual voice attendant product, using natural language processing and deep learning with python.

Entity capture (address, zip code, CPF, etc.) and connection to API via python. Work performed in squads (PO, SM, DEVS, devops, QA's, QM's) with

agile methodology and following the GitFlow flow. Tools used: Linux,

Python, Git, Github, bitbucket and MySQL. Deep learning frameworks:

Pytorch, Keras and Tensorflow.