Frederico Caixeta
AI/ML Engineer | AI Agents & LLM Systems
I build AI solutions that deliver measurable business impact—from achieving sub-3-second response times in banking chat interfaces to managing 1,000+ daily automated tests with 95%+ reliability in automotive ADAS validation.
About Me
AI/ML Engineer specializing in autonomous agents, RAG systems, and production-grade infrastructure that transforms how organizations extract intelligence from data.
I help teams build AI solutions that deliver measurable business impact through AI Agents with RAG/GraphRAG architectures, document intelligence through OCR and LLM integration, and cloud-native deployments on AWS.
Technical Focus
Python, Machine Learning, NLP
Graph & Vector Databases (Neo4j, Weaviate)
LLMs & Agent Protocols (MCP/A2A)
Cloud Infrastructure (AWS, Kubernetes)
Experience
Data Scientist
act digital
June 2025 - Present
Brasil
- ▸Developing AI Analysts Agents, RAG & GraphRAG products in banking background
- ▸Achieved sub-3-second time-to-first-token SLA for chat interfaces
- ▸Built RAG and GraphRAG solutions transforming unstructured banking data into actionable insights
- ▸Implemented agent-to-agent communication using MCP and A2A protocols
- ▸Scalable solutions on AWS (S3, Lambda, EKS) with Kubernetes-based vector storage
Founder
Authentic Developer
June 2024 - Present
Remote
- ▸Developing real-time sign language translation web application using Unity
- ▸ML models and LLMs power the translation pipeline for Libras (Brazilian Sign Language)
- ▸Designed to serve Brazil's 10+ million deaf and hard-of-hearing community
- ▸Scalable, browser-based accessibility solution for public spaces
Product Development Analyst
IAV do Brasil
June 2023 - June 2025
São Paulo, Brasil
- ▸Led mission-critical test infrastructure serving 200+ automated platforms with 95%+ reliability
- ▸Managed core Python software for ADAS validation testing, delivering 1,000+ tests daily
- ▸Created AI-driven scenario generation from natural language inputs
- ▸Developed ML models to predict test failures before execution
- ▸International coordination with teams in Spain and Germany
Featured Projects
Building AI solutions for social good, security, and enterprise transformation
AI Graph Analyst Agent for Barcelona Football Team
An Agent with direct access to neo4j graph database by MCP. The database has data coverage of passes that led to shots and goals. The Agent creates reports in a comprehensive way through natural language input, making odd cypher queries now democratic.
Content from Image Extraction with OCR-Tesseract
Extracting text from an image with OCR-Tesseract with a custom deskewing script. Also with a AI Agent connected to the extracted data, ready for a specific user query.
Deepfake Audio Detection
Machine learning models to identify AI-generated audio, helping combat misinformation and protect against audio deepfakes.
Fake News Detection Agent
AI-powered fact-checking agent that flags false information using LLMs and knowledge graphs to verify claims against trusted sources.
Real-Time Sign Language Translation
Web-based Unity application translating natural language to Libras (Brazilian Sign Language) through animated 3D characters.
Banking RAG & GraphRAG Systems
Enterprise AI analyst agents transforming unstructured banking data into actionable insights with sub-3-second response times.
Technical Skills
AI & Machine Learning
- •Large Language Models (LLMs)
- •RAG & GraphRAG Architectures
- •Natural Language Processing
- •AI Agents & MCP/A2A Protocols
- •Predictive ML Models
- •OCR & Document Intelligence
Infrastructure & Cloud
- •AWS (S3, Lambda, EKS)
- •Kubernetes
- •Vector Databases (Weaviate)
- •Graph Databases (Neo4j)
- •Docker
- •CI/CD Pipelines
Programming & Tools
- •Python
- •Machine Learning Frameworks
- •Tesseract OCR
- •Ollama
- •Unity (WebGL)
- •Git & Version Control
Specializations
- •Agent-to-Agent Communication
- •Real-Time Systems
- •Semantic Search (BM25)
- •Test Automation
- •International Team Coordination
- •Technical Documentation