NVIDIA Rubin: Six‑Chip AI Supercomputer and 10× Inference Token Cost Savings
Explore NVIDIA’s Rubin platform – a six‑chip rack‑scale supercomputer, its 10× lower inference token cost, and the observability tools that make large‑scale AI reliable.
Pillar two · Operations
Infrastructure & Reliability
Deployment, observability, cost control, self-hosting, and the unglamorous systems that keep products alive.
What belongs here
See the architecture and operational trade-offs behind reliable AI and data products.
7 published essays in this collection.
Self-Hosting Photos with Immich: An Architectural Deep Dive
A deep dive into self-hosting your photos and videos with Immich, exploring its architecture, features like semantic search and facial recognition, and backup strategies.
Architecting an Analytics Backend
This blog post details the architectural decisions and philosophies behind building a robust backend system for a startup, focusing on FastAPI, PostgreSQL, and advanced SQL techniques.
Architecting a Retention & Growth Platform Backend
This blog post details the architectural decisions and philosophies behind building a robust backend system for a retention and growth platform startup, focusing on FastAPI, PostgreSQL, and a modular design.
A Deployment Story: Taking a FastAPI AI Chatbot Live on DigitalOcean
How to deploy your fastapi application to Digital Ocean Droplet
Internals of the LangGraph Postgres Checkpointer (AsyncPostgresSaver)
An in-depth developer's guide to LangGraph's PostgreSQL checkpointer, its schema, and practical strategies for high-performance state management in LLM applications.
How to Add Supabase OAuth to FastAPI
A comprehensive guide on integrating Supabase OAuth with FastAPI for secure social logins using GitHub, Google, and other providers.