Adelson Teodoro
AI/ML Engineer
Technical Skills:
LLMs, Machine Learning, ML-Ops, Knowledge-Graphs, Computer Vision |(Python, TypeScript and Java)
Research interests:
1. Computer Vision, Text-To-Speech and Speech-To-Text, Languages.
Education
- ADS | Federal Institute of Tocantins (January 2023 - December 2025)
Work Experience
AI/ML Developer @ AutoU (feb 2025 - Present)
- Client @B3 - AI Developer Python, Docker, PyTorch, Transformers - Expertise in BERT (Imbau) and GPT models, Cloud solutions with Azure, Workflow management using Jira and ServiceNow
AI/ML Developer @ Compass.Uol (sep 2024 - feb 2025)
- Client @Tecban - AI Engineer - I specialized in building intelligent applications using GenAI (Bedrock, Claude Sonnet 3.5, Haiku 3) on AWS. My role involved:Serverless architecture with Lambdas, SQS, and Step Functions, Cloud infrastructure using AWS (S3, DynamoDB, Aurora Postgres); Infrastructure as Code with Terraform & Serverless Framework - MLOps & AI model deployment using PyTorch & Transformers, CI/CD and DevOps processes managed through Azure DevOps
AI/ML Developer with AWS @ Compass.Uol (Internship) (mar 2024 - aug 2024)
Development of systems with a strong focus on Machine Learning (ML) and Artificial Intelligence (AI). Experience in building and deploying scalable AI solutions, including:
- API development and Chatbots powered by AI
- Machine Learning models, covering Computer Vision and Natural Language Processing (NLP)
- MLOps pipelines for model training, deployment, and monitoring
- Cloud-native architectures leveraging AWS Cloud
- Serverless computing with AWS Lambda, Step Functions, and SQS
- Infrastructure as Code (IaC) using Terraform and Serverless Framework
Projects
LLM, NLP, Computer Vision & TTS
Trace Vision Chatbot
A system for registering, locating, and tracking missing persons using a chatbot. Users can report missing or found individuals by submitting photos and details.
Services:AWS Lambda, Amazon Bedrock, Amazon Polly, s3 Bucket, DynamoDB, Amazon Lex & Twilio Tools: Python & Serverless Framework
LaTEX AI
AI-powered tool that allows users to generate LaTeX documents from user inputs such as titles, content, themes, and images. The tool automatically creates structured LaTeX code for both articles and presentations, simplifying the creation of academic and professional documents.
Tools: Python (Flask), google-generativeai, Tex Live & Docker
Face Pet
The project aims to develop an API capable of recognizing people and their emotions through Amazon Rekognition, using the Serverless framework and receiving images manually uploaded to a bucket in Amazon S3. It is also possible to recognize pets and provide care, behavior, and health tips through Amazon Bedrock.
Tools: Python, Amazon Bedrock, Amazon Rekognition, S3 Bucket, Aws Lambda & Api Gateway
Eco Tech Bot
Chatbot developed using an LLM model with the aim of helping people correctly dispose of electronic waste in order to prevent environmental issues
Tools/Skills: Python, Docker, google-generativeai, few-shot prompt engineering, Flask Framework