58th Monthly Technical Session
Monthly Technical Session (MTS) is HENNGE’s mini-conference. As the name implies, MTS is held monthly and its talks are (mostly) about technology.
The 58th MTS was held on May 24th, 2019.
“From Zero to First Purchasing Order” by Xudong
Xudong recounted his experience working on one of HENNGE’s new services. This talk focuses on the period between February 2017 and January 2018, during which Xudong and others on his team started small, built prototypes, and continuously improved the service.
He promised to continue this story in the future MTSes, and we can’t wait.
“Enhancing Angklung Rehearsals with Technology” by Bumi
Angklung is a musical instrument from Indonesia made of a varying number of bamboo tubes attached to a bamboo frame. The sizes of the bamboo tubes determine the pitch of an angklung. In some cases, each performer in an angklung ensemble is responsible for several pitches. Performers sound their angklungs at the appropriate times to produce complete melodies.
Bumi set out to write a code that best distributes angklung in an ensemble. Angklungs have to be distributed in such a way that:
- No performer is responsible for multiple angklungs that has to be sounded at the same time
- Each performer gets a fair share of sounding their angklungs
- Each performer should be responsible for a well-proportioned group of angklungs
“Food for Thought in Your Ear” by Pete
A podcast is an episodic series of digital audio files which a user can download in order to listen, or as Pete explained it: radio-on-demand. It was born from radio programming, as their adaptation for the Internet era.
Apparently, today there are more than 700,000 active podcast titles available, with more than 29,000,000 episodes available. With such big numbers it is almost certain that podcasts cover a huge variety of contents, such as learning, journalism, storytelling, and many more. In this talk Pete shared some of his favorite podcasts — Stuff You Should Know, 99% Invisible, The Skeptics’ Guide to the Universe, and The Film Vault.
“CSS-in-JS” by Henry
For the longest time, ‘separation of concern’ is the recommended way to structure front-end code:
- HTML is used for organization of webpage content
- CSS is used for definition of content presentation style
- JavaScript defines how the content interacts and behaves with the user
Recently, component-based architecture is becoming more well-known. It suggests that tightly coupled parts can be encapsulated into independent, reusable components. There are several ways to style components: inline, using classes, using preprocessors, and — the one Henry recommended — using CSS-in-JS.
CSS-in-JS makes sense to Henry. According to him, ‘separation of concern’ does not necessarily mean ‘separation of technology’. He believes CSS-in-JS doubles down on the component-based architecture concepts — it encapsulates each responsibility into individual, modular, reusable components. Furthermore, with CSS-in-JS we can leverage the power of actual runtime scripts, we can remove classes (thus freeing us from one layer of abstraction), we can declare a template with built-in style, and we no longer need to assign props to classes. However, CSS-in-JS has a quite steep learning curve, its performance isn’t so good, and its code are not portable.
“Timelapse Video” by Billy
There are several ways of recording time-lapse videos. Some people use smartphone camera apps to do so, as most of them have time-lapse mode these days. Others capture lots of images and stitch them together afterwards to make a video. Billy recommended the latter approach.
Compared to recording time-lapse videos, capturing images allows for higher quality results. A normal video is recorded at 30 frames per second, which means the camera is capturing an image every 0.033 seconds. During low light conditions, 0.033 is simply too fast to let enough light enter the camera.
Creating good time-lapse videos require planning. Billy proposed several essential formulas you should prepare for the next time you are creating a time-lapse video:
- Number of images = frame rate * video length
- Period of capturing images = session length / number of images
For example, if you are planning to create a 10 seconds-long, 24 frames per second time-lapse video of a sunset from 5 PM to 7 PM (7200 seconds), then:
- You should capture 240 images
- You should capture an image every 30 seconds
“Can You Be Invisible Online?” by Xizi
Lots of stakeholders watches and listens to Internet users. Some of these stakeholders are definitely bad, such as malware and phishing websites. However, for better or worse, even stakeholders such as government security agencies and big companies observe people. For example, depending on the voice assistant app you use, they may listen to and keep all sorts of data. These data are used to train machine learning models, run ads, sold to third-party companies, etc.
In this talk Xizi explained other cases of people not being invisible online. Search engines collect data such as user agent, source IP address, location, and search queries. Browsers collect connection information, browsing history, mouse movement, device orientation, and others. Websites use the information provided by browsers to identify unique users and track their online behavior.
Fortunately Xizi also shared some countermeasures that we can use. DuckDuckGo is a search engine that apparently doesn’t collect or share personal information. You can use VPN for full message encryption and improved connection speed. Tor grants you nearly complete anonymity.
“Why You Should Care About Your Resting Heart Rate” by Yamato
Resting heart rate (RHR) is your heart rate while ‘at rest’. Most people are truly at rest moments after they wake up.
In general, the lower your RHR is the better. Researches show that higher RHR is linked with lower physical fitness, stress, anxiety, lack of sleep, overworking, excessive drinking, etc. People with RHR higher than 70 is twice as likely to die than those who are not.
Measuring this one single value allows you better understanding of your overall health. With gadgets such as smart watches and fitness trackers it is very easy to measure RHR. From Yamato’s own experience, measuring RHR allows him to better identify stress causes and motivates him to keep improving his habits.
“‘Not so Sexy’ Side of Data Science” by Din
Data scientist is arguably the sexiest job of the 21st century. There are high demands for this position and most of them pay well.
However, requirements for a data scientist position can be weird sometimes. One company may require you to do modeling on Microsoft Excel, while others ask you to do data warehousing.
The data science field has its own hierarchy of needs. From top to bottom, they are:
- Learn/optimize
- Aggregate/label
- Explore/transform
- Move/store
- Collect
Different companies have different interpretations of which needs should be accomodated by data science. Startups want data scientists to do all of those needs. Medium-sized companies want data scientists to do the first three. Large-sized companies want data scientists to do the first two.
Programming/hacking, math/statistics, and domain knowledge are very essential for data scientists. Different compositions of these attributes constitutes the various flavors of data science, for example:
- Machine learning engineers are good at programming and math
- Data analyst are good at math and domain knowledge
- Data engineer are good at programming and domain knowledge
- Machine learning researchers are very good at math
- Analysts are very good at domain knowledge
- DevOps are very good at programming
As usual, we had a party afterwards :)
