AI engineering for developers

AI engineering for developers
about 1 hour by Luca Cavallin

Luca introduces AI engineering for software developers moving into building AI-powered applications. He explains foundation models, prompting, RAG, agents, evaluation, finetuning, and inference in simple, practical terms. The focus is on real production challenges such as cost, reliability, scalability, and maintenance. Using modern tools and examples, Luca shows how engineers can design, deploy, and manage AI systems effectively without needing deep machine learning expertise.

Tired of DocuSign API complexity and high costs?
sponsored by BoldSign

BoldSign is a secure, developer-friendly e-signature platform for embedding signing workflows directly into applications. With its REST eSignature API and SDKs, you can send signature requests, track status in real time, and retrieve signed documents. It also supports branded signing experiences for a seamless user journey. With transparent and affordable pricing, BoldSign is a strong alternative to Docusign.

The invisible engineering behind Lambda’s network
14 minutes by Werner Vogels

AWS Lambda's networking team spent nearly a decade rebuilding the system that connects serverless functions to private networks. They cut tunnel setup time from 150 milliseconds to 200 microseconds using eBPF, replaced stateful NAT with stateless packet rewriting, and reduced root namespace firewall rules from 125,000 to 144. The result: a single unified network topology supporting 4,000 virtual machines per worker, 20 times more than before, while using less CPU.

Selective test execution at Stripe: Fast CI for a 50M-line ruby monorepo
11 minutes by Aditya Anchuri

Stripe runs about 100,000 Ruby test files that would take four months to complete sequentially. To stay fast, their system tracks which files each test actually opens at runtime, using a lightweight C++ library, then only reruns tests touching changed files. This cuts the average test run to about 5% of the full suite. A compact bitmap index and a simple database make lookups fast and repeatable across roughly 50,000 weekly builds.

Being oncall taught me everything
4 minutes by Yao Yue

Yao reflects on how years of being oncall shaped her as an infrastructure engineer. Through handling large-scale incidents at Twitter, Yao learned key technical lessons about reliability, system design, and operational excellence, while also developing accountability, teamwork, and resilience. She argues that real understanding of software comes from operating and debugging systems in production, where collaboration and responsibility matter as much as technical skill.

Things you didn't know about indexes
11 minutes by Jon Charter

Database indexes work like a textbook index, letting the database jump straight to data instead of reading every row. Adding indexes speeds up reads but slows down writes and uses extra storage, so they need careful thought. Common mistakes include column order in multi-column indexes and wrapping columns in functions, both of which can stop indexes from being used. Tools like EXPLAIN help you see exactly what the database is doing.

And the most popular article from the last issue was:

• A history of IDEs at Google