___luigi

In the ML industry, there are two type of companies.

1/ Open source companies like huggingface (https://github.com/huggingface), explosion (https://github.com/explosion), Fast.ai that democratized access to ML and provided a set of tools & language models for engineers. These companies didn't only manage to build a company and open source projects, they built a welcoming community and it's impressive to see all these students use these tools to tackle bigger problems. That's what"democratized" mean to an engineer[1].

2/ Media companies that built ML APIs to ease the use of such services like OpenAI. I think this company falls under this category.

[1]: https://marksaroufim.substack.com/p/machine-learning-the-gre...

Ask your employer, most people don't know about it.

> .. I see have almost zero history of contributing to OSS or maintaining side projects.

Not sure about US, but in EU, the contract from employers doesn't allow you to contribute code without permission (e.g. contributing to OSS, working on side projects, etc), it's very very rare to find an exception.

You can check this discussion few weeks ago: https://news.ycombinator.com/item?id=27843198

> I have to rely on that ..

By doing this, you are not considering employees who are doing nothing wrong, but follow the rules and respect the contract terms.

GREAT work!

I can understand why big companies use LC type of questions to assess coding skills of those who graduated recently or didn't enough experience to show their work, but I struggle to understand why many Startups adopted such way of evaluating engineers. I see many startups nowadays asking to candidates to reverse a tree, a task that doesn't reflect the daily work in a startup (where people change gears frequently). I understand that people interview to join the big company, pass a certain bar and then get matched to a team, the startup is too small to not have clarity on what skillset they are looking for.

> .. scaling

Did you try Stolon?. https://github.com/sorintlab/stolon

Is it better in neighboring countries?. There was a discussion [1] about controlled rent market few days ago, it seems that the problem is related to growth.

[1] https://www.bbc.com/news/business-58317555

> walking back to hotel at night, only about 30% of houses had any lights on.

I wish a data scientist at utilities (water, electricity and gas) providers pick up this line and provide some statistics on this. I wish also that city-services companies (e.g. Electricity/water/parking service provider) can collaborate to provide better and justified policy making.

I did similar studies based on open data to provide insights on road/sidewalk optimization usage, and I still see value in collaboration with policy makers.

I deleted my Linkedin account as well. I use other social platforms.

I think you can use your email to signup. Linkedin is just an option.

What do you mean by "optional"?. You are including it [2] [3] [4] as one of the options (not re-implementation of the original paper [1]) and it is one of the dependencies [4] (Prophet [5] is an open-source library built by facebook research).

I think we should distinguish between science/open-source and policies when we mention open-source projects.

*References*: [1]: https://facebook.github.io/prophet/#:~:text=Prophet%20is%20a....

[2]: https://github.com/unit8co/darts/blob/master/darts/models/pr...

[3]: https://github.com/unit8co/darts/blob/master/setup.py#L12

[4]: https://github.com/unit8co/darts/blob/master/requirements/pr...

[5]: https://github.com/facebook/prophet

> .. bigger echo chamber than FB.

I would question this line, FB has a serious challenge to address in this space.

I'd like to hear responses from those who work at big companies like Facebook. The media painted a picture that the company is putting no efforts on "managing" political divid or fighting misinformation. As researcher/engineer, I can see that FB/Google spend a lot of money in research in this area (I worked in this area for some time), but I have no clue how does this translate in practice and deploy it in production?. It's a borderline also when someone works in a company where it does make efforts to fight a problem, but it is not succeeding at solving the issue, and work for a company that avoid solving the problem.

We were thinking to switch to using Notion for a sizable project, but this thread changed our mind. It is, indeed, frustrating.

This is one of those timeless HN comments that impact any parent's life (in a positive sense).

Yes, many companies did that. In our company, we were allowed to take any device/desk to home.

In Switzerland, Gov asked companies to pay % of rent. https://www.swissinfo.ch/eng/court-decision_companies-must-p...

The only exception was Germany. There is ongoing discussion that Remote workers should pay a (*5% tax*) for having flexibility/option to work from home. https://www.cnbc.com/2020/11/12/deutsche-bank-proposes-a-5pe...

I'm not a teacher, but I heard good things about replit https://replit.com/teams-for-education

> .. but the inefficiencies of technology .. can't always be resolved by technology alone

Technology will always resolve inefficiencies of technology. Check your mobile in your pocket, and mobile in the 90's.

I am a bit biased [1] [2], I work on ML. That was my naive observation.

[1] https://www.natureindex.com/news-blog/google-scholar-reveals...

[2] https://www.natureindex.com/news-blog/google-scholar-reveals...

A good catch. I was typing fast, and I didn't pay attention to it.

> .. may one day pose similar risks due to the run-away prestige games scientists play.

It is happening in front of our eyes. From where should we start.

(1) Responsible AI, but the outcomes didn't challenge AI research enough to stop some of the practices. I would avoid commenting on what happened at Google AI recently, but I would say that most of the AI-based solutions that are deployed at large scale can be challenged.

(2) paperswithcode.com solved this issue. Back in the day, we would re-implement models that we see in the paper. When we email authors, we will always figure that an RA implemented the solution and s/he left.

(4) Rich is becoming richer. There is a phenomena in academia to blindly trust those who are in good labs. I have seen many papers that are great coming from these labs, but there are bad papers as well. we shouldn't just blindly trust ppl.

(5) Review process is broken. Review is a voluntarily work for professors, and imho it shouldn't. In industrial R&D, reviews are one of the most important things as part of the work. Some of our PRs takes days.

(6) All recommendation engines are black-boxes. I don't know what YT or FB recommend to a kid and whether this is aligned with some of our values. If s/he accidentally watch something bad, we noticed that their timeline is full by the same ideas. In our spare time, we tried to check if we can influence this behavior as a user. I figured that all of our recommendations engines are black-boxes. I don't know why we don't challenges. I hacked an App to fetch data, new, feed and run it through recommendation engine and started customizing what I can read.

(7) It's 2021, most of good papers are coming from industrial labs.

You should check bitcoin early-days posts

No. G+ was tech-focused social platform, dominated by Google-rs.

I come from no-name university in the global south, and network/friends is the very important thing. Literally, I have many friends who traveled the world, worked in different places, and then came back to start a business together (with homies).

> That statement is true for a small set of elite universities.

In fact, elite universities tend to have good program and the prestige can take to places.

Universities are not about classes, it's about network out-side the classes. We all know that from experience, but Covid shaped us as a society in different ways. Companies will become global. Being a digital nomad will be normalized, and the difference between an entrepreneurs will be about ecosystem and entrepreneur status in the country.

Network effects will be always there, but not as before.

Stackoverflow is a very toxic community. It's not a welcoming place for beginners who are basically want to learn how to learn. An alternative was to use discuss forums (e.g. https://forums.fast.ai/).

I have spent time responding to questions on Stackoverflow, I think I spent 2-3 years providing answeers, and one time I answered a question that didn't follow a weird policy that I strongly don't agree with (https://meta.stackoverflow.com/questions/311442/opinion-base...). I got banned, and my account has been deleted, the answers still there. My only regret was that I could have spent that energy and drive to contribute to open source projects on github. Now, I have less time, but I answer questions -when I have time- on reddit or forums.

Would you consider HuggingFace as as industrial research group?.

That's impressive. This reminds me of a story I read about being an independent researcher (https://andreas-madsen.medium.com/becoming-an-independent-re...).

For those who want to do the same. When did you start?. How did you start?. What kind of setup did you have?. How did you stay focus?, and for how long did you work on that?. Any failure stories you want to share?.

>> .. a "no-name" german university

We come from different places. German universities are good in average. + MSc student have the chance to work as "working students" (20hr/week) at any company in Germany. https://www.topuniversities.com/university-rankings/world-un...

>> .. It's just hard to publish in those top-tier conferences

Hard is relative. You should also check the costs of publishing a paper or attending a conference for students from global south https://twitter.com/hadyelsahar/status/1374699909451030542

>> But you can do a phd here without publishing a paper first

I agree with that.

> .. more money most obviously of all.

I would challenge this point. There are few research labs in Singapore and middle east (UAE, Saudi Arabia) where money is not a problem. I have seen many researchers fly there to build research groups, but many of them leave after some time. You can see papers coming from these groups in top-tier conferences (NeurIPS, KDD, etc) every years, but these papers come from individuals, not groups. Money is one ingredient, you need fresh blood and students to work on new ideas. You need an ecosystem for research and industry to adopt and develop some of these ideas (e.g. Standford/Google, UCLA/DeepMind, UoT/Uber, MIT/Boston startups, etc)

Sorry to bring the bad news, but the author the first comment is 100% right. Moreover, structure in academia is different than in Startups. In startups, people don't care about "names" and "prestige". In academia, it is different. In order to get your next job in academia, you need (1) recommendation letters (2) citations. Advisors usually have a larger network and experience to be able to publish papers and help you climb the ladder through their networks/recommendations.

In Industry, you can come from no-name university, and you can still make it to FAANG (or top tier companies). In academia, that's impossible. To start, it is impossible to make it to good PhD program without publishing papers in top-tier conference, MSc student in no-name universities can't afford finding advisor to easily do that. Good PhD programs will help you to join a good Lab, and usually it is through network + recommendations.

In Startup world, you can sit in your dorm room, without anyone believing in you and you can make it.

P.S. I am a research engineer, I worked in both places and that was my observations. There is no shortcuts, everything is hierarchical in academia.