Implement Night Mode on Twitter Lite

We have just launched Night Mode on Twitter Lite recently. Night mode is an exciting feature in regards to engineering. It is a highly demanded, visually pleasing and the primary driver for our effort of moving our CSS to CSS-in-JS. Let’s dive into what did we do to bring this feature to life!

Infinite List and React

I have worked on Twitter’s new mobile website for the past year. We rebuilt the website using the latest web technologies: React, Redux, Node.js/Express to name a few. It is absolutely an exciting project to work on since you rarely get a chance to rework a large-scale website from the ground up and experiment with the latest tools without having to worry about any historical baggage.

OpenSesame: Reverse Engineering BLE Lock

Open Sesame. Sesame is a smart door lock from the CandyHouse. It uses Bluetooth Low Energy to communicate wirelessly with smartphone apps. We are going to explain its BLE protocol and how we can write a script to control it. The protocol is reverse engineered from its Android app. This is not a full protocol documentation. I only reversed it just enough to lock/unlock the door.

Introduction to Twitter's Zipkin

Zipkin is the Twitter open-source implementation of Google’s distributed tracing system, Dapper. It’s a great tool for people who wants to understand the bottleneck in their multi-services system. The only downside is that I found its documentation isn’t quiet clear about the tracing format, so I decided to write a blog post that gives an overview of the system and its communication protocol.

Two-factor Authentication for Windows

I had a remote running Windows with Remote Desktop enabled. I’m really uncomfortable about just putting it online with the bare minimal Windows authentication protection, so I went online to do a little research and I found out that the algorithm Google Authenticator is using (it’s called TOTP) could be integrated to Windows authentication system. Here is how to do it.

(my way to) Write a Hangman AI

I was asked to write a hangman AI as a challenge last week. I was asked not to leak the detail, so I will not post my solution here. However, the hangman problem itself is a well-known problem, I would like to share some thoughts that I used to build my hangman solver.