portfolio.tsx

Leonardo Avanzi

Building things people actually use. When not coding, you'll find me swimming or pretending I know what I'm doing at the gym.

GitHubLinkedInEmail

profile.json

Leonardo Avanzi

status

shipping

mode

engineering

country

🇧🇷 Brazil

New article

Useful tips for running open-source models on Google Colab

A practical look at GPU reuse, xFormers, dependencies, and video pipeline bottlenecks.

Read now ->

colab-pipeline.gif

AI video pipeline preview

workspace

main.content

blog.mdx

Loading articles...

stack

Tech Stack

Technologies I've worked with

tech-stack.json

Languages

Java
TypeScript
Golang
Python
Elixir

Backend

Quarkus
Phoenix
Node.js
NestJS
Hono
Spring Boot
Zod
WebSockets

Frontend

React
Next.js
Angular

Data & Search

PostgreSQL
pgvector
MongoDB
Redis
Elasticsearch

Cloud

AWS
GCP
Vertex AI
Compute Engine
Cloud Run
Cloud Functions (v1 & v2)
Cloud TTS
Artifact Registry
BigQuery
Pub/Sub
Firebase
Firestore
Railway

DevOps & Infra

Docker
Kubernetes
GitHub Actions
Bun
Datadog
Grafana

Media Processing

FFmpeg
Puppeteer

AI & LLMs

Claude
OpenAI
OpenAI Embeddings
Whisper
Gemini
LangChain
RAG
LocalAI
HeyGen
ElevenLabs
MiniMax
Grok API
Hugging Face

Machine Learning

TensorFlow
Scikit-learn
OpenCV
MLflow
NLP
Computer Vision
Recommendation Systems