charlie0077

hmm, thanks for the input

Hmm, thanks for the info. Just to clarify, it will be appointment based, and you will go to a meeting link similar to zoom meeting. They will never call you directly, neither do they know your contact info. You can cancel the appointment anytime if you don't like talking to that person.

A lot of people in Asian countries can read, listen, write, but not able to speak. They don't have the environment to practice.

Hi, thanks for all your comments.

All the things you mentioned totally make sense. I do have a landing page. This is to quickly see the other people's thoughts and validate the idea.

Eventually it would be topic based and/or algorithm driven. But for now, it would be manual, it just takes time and effort to build everything. I am doing this for free and for fun.

I have not thought about the abuse issue, looks like a pretty big and valid concern.

I notice that Hasura does not support groupBy. And I don't like a blackbox service. I eventually built my own library: https://github.com/charlie0077/graphql-server-crud/ I am curious to see your thoughts about it.

Why not node.js instead of Django? In general I feel node.js community is more active in terms of web development. And you don't need to do context switch between node and python. I would go with GraphQL, the part that I like the most is the schema. You don't need to maintain doc now. There is also a playground, you can do integration test easily during development.

If you feel bad about all the boilerplate code you need to write in server side. There are many node libs/blackbox services that can help you, such as prisma, hasura. You can also take a look at a lightweight library that makes your life easier when building GraphQL server: https://github.com/charlie0077/graphql-server-crud/

I am the author of that library.

Thanks for sharing. I totally feel you. I spent long time doing research and not super happy about what I found. I would say prisma/hasura is pretty good at this point, however, as a "minimalist", I just feel it is better to have some light weight "helpers" that does most the jobs and give you full control of everything, no extra process, no new things to learn, no blackbox, etc.

I am not a ML expert, so I cannot give you correct guidelines. But I do know that there are a lot of problems in kaggle. You can pick a direction/domain: cv, reinforce learning, nn, etc.

Personal opinion to your 2 problems: 1. Work with your teammates to figure out what is the best for the company. It may be hard, due to communication, personality, etc. But the bright side is now you get a chance to show your skill, your leadership. Maybe you are able to be the lead of the engineering team after 2 years, and become a partner of the company owner. If you feel the senior is making bad decisions frequently, talk to them, talk to their boss, talk to the company owner, WITH PROVE. (if the company's culture is truly toxic, quitting the job is also a good option).

2. don't worry about test. the most popular open source project(linux kernel) do not have test code. it is overrated. When you interview for a job, nobody will ask you how good you are in writing tests. If I were you, I would prefer no test code required, so that I can focus on learning more important stuffs to grow myself: design pattern, infrastructure consideration.

3. about ML: you don't need degree. if you can be top 10 in kaggle, nobody cares your degree. If you truly like ML, you probably don't care about the job. How do you learn ML: google, online courses, and the most important: projects. The best case is that you can work for a company that is willing to hire you as a software engineer but then they allow you to work on ML projects. But this is kind of luck and you have to do google search to find that type of company. But, getting a degree and naturally getting hired as a ML engineer would be safer plan if you have decided ML is what you like. But there is cost: time and money.

You should try it if you don't like black box.

I feel the same pain and end up built a library to try to solve it. https://github.com/charlie0077/graphql-server-crud