Software Engineer | Backend, Applied AI & Cloud

I'm an Information Systems undergraduate at the Federal University of Rio Grande (FURG), with expected graduation in 2027, and work on applied R&D projects at iTec/FURG's EMBRAPII unit.

I build backend systems, applied machine learning solutions, and cloud infrastructure, primarily with Python, Go, AWS, and Terraform. My interests span the full engineering lifecycle: designing APIs and data pipelines, integrating ML models, automating deployment, and operating the resulting systems.

A self-hosted AI document processing platform built for resource-constrained, CPU-only infrastructure. A Go/Echo API gateway coordinates asynchronous Python workers through Redis to provide document summarization, multilingual RAG, and speech transcription using locally hosted models.

The system includes Qwen3 inference via llama.cpp, FAISS/FastEmbed retrieval, Faster-Whisper, S3-compatible storage (MinIO/Cloudflare R2), SSE token streaming, and content-addressed deduplication. I also implemented isolated Go/Python tests and CI workflows around the processing pipeline.

Live application →

Go · Python · Redis · Qwen3 · llama.cpp · FAISS · Docker · Cloudflare R2 · Next.js

An end-to-end movie recommendation system developed using 4.3M+ Letterboxd interactions. Its candidate-generation pipeline combines TruncatedSVD collaborative filtering, FAISS retrieval, and a popularity baseline, with offline experiments used to evaluate retrieval and ranking decisions.

Historical offline evaluation reported 84.3% candidate recall and 99.8% catalog coverage; these are candidate-retrieval metrics, not measures of online user satisfaction. Beyond modeling, the system provides FastAPI/Next.js serving, asynchronous profile ingestion with Celery/Redis, and a versioned model lifecycle with validation, promotion, and rollback.

Live application →

Python · FastAPI · PostgreSQL · scikit-learn · FAISS · Celery · Redis · Next.js

An automated newsletter and web dashboard that gathers content from 14 AI and software engineering sources on weekdays. A Go-based AWS Lambda pipeline, scheduled through EventBridge, evaluates and summarizes articles with Amazon Bedrock, stores results in DynamoDB, and distributes editions through SES.

The frontend is delivered through S3 and CloudFront, with infrastructure defined using AWS SAM and deployments automated through GitHub Actions. Estimated LLM inference spending is approximately US$2/month (not total AWS operating cost).

Live dashboard →

Go · AWS Lambda · EventBridge · Bedrock · DynamoDB · SES · S3 · CloudFront · AWS SAM

A reproducible infrastructure stack for running multiple applications on a small Debian VPS. It provides Caddy reverse proxy and automatic HTTPS, isolated Docker networks, shared PostgreSQL/Redis services, Prometheus/Grafana monitoring, and automated PostgreSQL backups to off-server object storage.

The repository separates application deployment from shared infrastructure management and includes bootstrap, health-check, backup, and restoration workflows.

Linux · Docker Compose · Caddy · PostgreSQL · Redis · Prometheus · Grafana · Shell

At iTec/FURG, I contribute to two applied R&D projects:

• CELIA - Fleet Optimization: Primary developer of the backend and a genetic-algorithm optimization engine. I implement multi-objective fitness functions involving distance, time, and operational cost, subject to constraints such as vehicle capacity and service windows, and expose scenario results through an API backed by PostgreSQL.
• LAPIS - AWS Infrastructure & ML Systems: I develop reusable Terraform modules and cloud deployment orchestration. The project's third architecture scenario uses an event-driven workflow (SQS → EventBridge Pipes → Step Functions) integrating ECS Fargate, Lambda, SageMaker asynchronous inference (Qwen3), SageMaker serverless inference (XGBoost), Bedrock, Aurora PostgreSQL, and OpenSearch. My contributions focus on infrastructure as code, service integration, and deployment automation.

Previously, during the BRISA ICT Residency, I designed and implemented the backend of a psychology clinic's administrative system using Django REST Framework and PostgreSQL, including authentication, appointments, financial workflows, reporting, and audit functionality. The application was deployed and used by the clinic.

• Languages: Python, Go, TypeScript, SQL
• Machine Learning & AI: PyTorch, scikit-learn, LightGBM, NumPy, Pandas, FAISS, Recommender Systems, LLMs, RAG, YOLO, genetic algorithms
• Cloud & Infrastructure: AWS (EC2, ECS/Fargate, Lambda, S3, RDS/Aurora, SageMaker, Bedrock, SQS, EventBridge, Step Functions), Terraform, Docker, Linux, GitHub Actions, CI/CD
• Backend & Data: FastAPI, Django REST Framework, Echo, PostgreSQL, Redis, SQLAlchemy, Alembic, REST APIs, OpenAPI
• Observability & Systems: Prometheus, Grafana, event-driven architecture, asynchronous processing, serverless, infrastructure as code

I'm interested in engineering roles at the intersection of software development, applied AI/ML, and cloud infrastructure. I particularly enjoy building complete systems and understanding how architectural decisions affect reliability, performance, and cost.

Based in Rio Grande, Brazil, and open to remote junior and internship opportunities. I speak Portuguese (native) and English (fluent).

Outside engineering, I'm interested in cinema, chess, and music.

LinkedIn · Email