HNHacker News
TopNewBestAskShowJobs

baristaGeek

956 karma · joined June 17, 2014

Software Engineer at Kiva.org

https://twitter.com/helloteban https://github.com/baristaGeek https://www.linkedin.com/in/tebanvargas/

Feel free to ping me! estebanvargas94 [at] gmail [dot] com

[ my public key: https://keybase.io/baristageek; my proof: https://keybase.io/baristageek/sigs/xvrFueBx0GMNM1qPjertwxrGmgyPzGzabPoXl_Q02Io ]

Medellin, Colombia

submissionscomments
baristaGeek··on McClure steps back at 500Startups after internal sexual misconduct investigation
How can a newly born organization build an organizational culture that avoids sexual misconduct as much as possible?
baristaGeek··on Tensorflow v1.2 released
Tensorflow implements the (really abstract, at least for me) concept of a computational graph. Think of it as nodes being values and edges operations.

Among other things, what this allows you to do is partition your computational graph into different subgraphs and run each subgraph on parallel.

Sklearn doesn't allow you to run things on parallel; however, I do agree that TF doesn't have a favorable learning curve, so you might want to start with SKlearn (or TFLearn) to get to know the basics of ML first.

baristaGeek··on A novel approach to neural machine translation
That's a great explanation.

Let me add this though:

Artificial neural networks were proposed to compute the probability of a sequence of words occurring; however, RNNs were the next step in Natural Language Processing since they allow variable-length sequences to be received as an input contrary to the previously proposed architecture.

However a simple RNN architecture didn't allow for long -term dependencies to be captured (that is, use statistical modeling to predict a word sequence on a part of a text that is based on an idea previously developed on the corpus). So two kinds of fancy RNN architectures were developed to tackle this problem: GRUs and LSTMs. Production systems are already implementing these architectures and they are yielding pretty accurate results.

But now Facebook researchers are proposing using CNNs for this task because this architecture can take more advantage of GPU parallelism.

baristaGeek··on YC will hold interviews in Vancouver for founders who can’t get US visas
Isn't this going against the recently approved International Entrepreneur Rule?

EDIT: It'll work pretty well for people coming from countries which have the Muslim ban.

baristaGeek··on Numbers 0 to 11111 in terms of Increasing and Decreasing Orders of 1 to 9 (2014)
Reminds me so much of competitive programming
baristaGeek··on H-1B Visas: U.S. Lawmaker Re-Introduces Bill to Tighten Rules
Isn't it a HUGE problem that it's not discriminated by cities according to its living costs?

I'm not an expert in macroeconomic policy but I guess it wouldn't be so hard ad doing the discrimination for say the minimum salary, as business owners/shareholders tend to have a better understanding of the concept of the price index than minimum salary earners.

And I'm saying this because tech companies in the Bay Area won't have such a hard time following this rule than startups in cheaper cities.

Which would be a recoil in terms of making all America have knowledge-based industries. Wouldn't it?

As for people taking the H-1B, would this be something better? Or worse?

baristaGeek··on Ask HN: What do you want to learn in 2017?
1) Deep Learning 2) Figure my way around DevOps 3) Get more proficienet with ES6 and Go
baristaGeek··on Show HN: The Road to learn React – Build a Hacker News App on the Way
Why would you want to teach React without Redux?

I failed on the trap as well, I mean, why would you want to add yet another 'concept' to your stack when you can just get a simpler piece of code do the same thing as a more complex one?

The truth is that if you're going to build something as simple as a simplified Hacker News clone, you can just get along using concepts of Redux with plain-old React without actually using Redux.

However, if you want to build a much more complex frontend system using Redux does give you a utility.

baristaGeek··on Venezuela’s Currency Has Had its Biggest Monthly Collapse
Actually Venezuela used to be my country's (Colombia) closest trade partner. Today it's the US, but our exports have definitely been affected because Maduro can't stop printing bills just as a kid who gets prohibited eating candy can't stop eating candy.

I'm also against interventionism in general, but Venezuela needs it.

baristaGeek··on React component to PNG converter
Nice!

It would be good if you added some contribution guidelines :)

baristaGeek··on Donald Trump Is Elected President
Just in case anyone is wondering the same thing as I last night:

The USCIS administrative act for the International Entrepreneur Rule (Startup Visa) won't be affected. First, because it's being passed as an administrative action, meaning that it doesn't have to be voted in congress in order for it to start taking an effect. Second, because it's actually aligned with Donald Trump's 3rd immigration policy driver (the first two have nothing to do with foreign-born entrepreneurs):

3. A nation that does not serve its own citizens is not a nation. Any immigration plan must improve jobs, wages and security for all Americans.

What Trump does believe is that the H1B1 is depressing salaries thus decreasing job opportunities for american citizens.

Therefore, if you have a startup and want to move it to the US you will have a better pathway. However, if you want to work as an employee in America, you'll have an even harder time.

baristaGeek··on Overview of JavaScript ES6 features
Have used it in production and it's amazing.

Have you tried Minify(Babili) by the way?

baristaGeek··on Overview of JavaScript ES6 features
If you're developing a desktop-only app (or you're making a native app for the mobile frontend) ES6 works pretty well.

At least I haven't received any complaint yet, and if you can optimize early on without any short-term cost then it's the best thing you can do.

Thank god Chrome and Firefox dominate the market.

baristaGeek··on RethinkDB is shutting down
I have tinkered with RethinkDB a lot recently. It's the NoSQL DB (and the DB of any kind, I believe) that has the deepest focus on real-time features, that mixed up with Golang's concurrent model can support really powerful and highly scalable real-time apps.

But my question is, what are your thoughts on building more than a product but a business with RethinkDB? Will the support quality decrease? Will Stripe offer architectural consulting? Will developers be interested in learning ReQL?

baristaGeek··on Colombian Voters Reject Peace Deal with Rebels
This is the 'tropical' version of the Brexit: Old and uneducated people ruined the country for young people.
baristaGeek··on Colombian Voters Reject Peace Deal with Rebels
The 297 pages of the agreement have a lot of juridical language that's really hard for non-eduacted people to understand. However; magazines, newspapers, tv commercials, YouTube channels, etc. made huge efforts to simplify and summarize the document for the average-joe.

What's really interesting is that people in the major cities (where we're alien from war) voted to reject the agreement. In the marginal zones (where conflict occured/occurs) people voted to accept the agreement. The direct victims just want their towns to stop being bombed all the time, if FARC leaders get money or seats in the congress that is just a minor detail for them.

If the government didn't care about people's opinion they would've just imposed the agreements and jump the plebiscite. I agree with you that the celebration shown in the news was political propaganda for the 'YES', but in no way that's passing over people's opinion.

You don't want to tell your children that you have congressmen who commited many of the worse war crimes? In which country do you think we live in? This is already true, so adding FARC members to the congress in no way changes that. Also, FARC was only going to get 5 out of 102 seats in the senate and 5 out of 166 seats in the chamber of representatives. When M-19 and other rebel groups got their seats in the senate back in '91 they got 15 seats, and today we're not a socialist state. Also, compare this peace process with the ones in countries such as Nepal, Nigeria or Sudan. I mean, in Sierra Leon the rebels got the vicepresdiency, 4 ministries and 4 viceministries.

You talk about brain drain in Colombia. The 'YES' won by a huge difference in the voting tables overseas. Following the logic that says that geniuses leave the country (which I believe it's true by the way), geniuses voted 'YES'.

What's gonna happen next? Well, first of all we will see if FARC really has intentions of building peace or not.

Second, an 'asamblea constituyente' will be built. This will not only prolong the peace process, but will ironically give more negotiating power to FARC.

baristaGeek··on How Should We Use Entropy in Economics? (1991) [pdf]
> Presumably, a process of trial and error will be needed to find the right choices

This is a huge difference between macroeconomics and thermodynamics. You can't just have a control group of people, test a policy with them and say 'omg they died of starvation'. I know well designed experiments exist such as basic income, but I hope I made a point: You can't just test macro policies as you would test heat transfer in a lab.

baristaGeek··on Go success stories from around the web
I've been using Go for 4 months now and I'm loving it.

The syntax is elegant, concurrencies consume an order of magnitude less in the memory stack and it has such a good performance benchmarking.

What excites me the most though is the TensorFlow wrapper for Go that Google has recently stated it's developing. It's a huge advance towards high-scale, real-time deep learning.

BTW, in a few days I'm giving an introductory workshop to backend development using Go, I'll talk about some of the examples mentioned for sure.

However, I know that Dropbox is one of the biggest Go users and it isn't in this success stories list. Anyone knows more about it?

baristaGeek··on Ask HN: What are the must-read books about economics/finance?
People have mentioned different authors across different schools of economic thoughts such as Mankiw, Rothbard, Friedman, Hayek, Smith, Keynes, etc. There's one that's also being mentioned which I particularly would avoid recommending which is Piketty.

Those are the best recommendations.

I would like to give a recommendation that might be a little bit different: 'Why Nations Fail' by Acemoglu.

baristaGeek··on Ask HN: What are the must-read books about economics/finance?
Mankiw's books is the best option if you want an inferential and qualitative understanding of the subject. If you want something with more mathematical rigor I suggest Blanchard's Macroeconomics.
baristaGeek··on Why building a startup is probably your most sensible career path
"Now, the most successful investors are increasingly coming to the realisation that searching for the low hanging fruit in web-tech is nearing the point of diminishing returns."

This is true in the sense that the amount of VC poured into things such as social networks or SoLoMo startups is decreasing and the amount of VC poured into things such as machine-learning based startups is increasing. Even though the amount poured into the first is still larger than the amount poured into the second.

However, this article shouldn't discourage people with just an undergrad (or even without an undergrad) to start startups. A group of 22 year olds might know about Deep Learning and might just apply it as a medium to a solution in a different narrow domain than a group of 40 year olds will.

baristaGeek··on Machine Learning in a Year
Can anyone tell us how hands-on is Stanford's CS224D please? I found Andrew Ng's course too theoretical and not so practical.

Right now I'm almost halfway through Udacity's Intro to Data Science and it's great, so I'm deciding if I should take Udacity's Deep Learning or CS224D

baristaGeek··on How to Become a C.E.O.? The Quickest Path Is a Winding One
Actually your chances are increased if you come from the middle class. I might get downvoted for writing this, but the poor have a harder time because of their lack of access to education and the rich have it 'harder' in the sense that they don't have certain motivations to work harder. These are tendencies of course, not the general case.

I remember reading the Forbes 30 under 30s stats once and seeing that >60% came from a middle-class family.

I'm just 22 years old but from what I've seen for the moment from people around my life, this is what happens: Guys with really rich parents end up at an important role at their dad's company; and once their dad retires they become CEO. On the other side, middle-class guys are the ones starting up startups and companies in general; which ends up making the number of middle-class CEOs superior to the number of high-class CEOs.

baristaGeek··on Regular Expression That Checks If A Number Is Prime
Performing a Miller-Rabin primality test or filling a Sieve of Eratosthenes will almost always be the way to go. However, this line of code seems like magic and that's kind of impressive.
baristaGeek··on Regular Expression That Checks If A Number Is Prime
Filling the Sieve has a time complexity of O(n log log n), and accessing a value is clearly O(1). This is far away from that.
baristaGeek··on Google and Box Announce Partnership
IPO once and for all?
baristaGeek··on Why is printing “B” dramatically slower than printing “#”? (2014)
Let's talk about any potential application this could have.

The first thing I can think of is that if you're writing a flood-filling algorithm (finding connected components) you should fill with #s instaed of Bs, even if filling with Bs is more conventional. Specially if you're doing competitive programming and need to overkill the optimization process in order to pass the run time.

As for production coding, I can't think of anything. Any ideas?

BTW, this is probably speculation, but it's still worth discussing in my humble opinion.

baristaGeek··on Shakespeare's language not as original as dictionaries think
Indeed. I'm just summarizing what I read. Nothing uncanny about it :)
baristaGeek··on Shakespeare's language not as original as dictionaries think
Plutarch wrote 'Lives of the Noble Greeks and Romans' (also called Parallel Lives or Plutarch's Lives) somewhere around the second century AD and Shakespeare wrote 'Antony and Cleopatra' in 1606. Shakespeare took 'some' inspiration from Plutarch. Plutarch wrote in Greek, Shakespeare in Elizabethian English.

Here's Thomas North's translation of Plutarch's work:

"...she disdained to set forward otherwise but to take her barge in the River Cydnus, the poop whereof was of gold, the sails of purple, and the oars of silver, which kept stroke in rowing after the sound of the music of flutes, howboys, cithernes, viols, and such other instruments as they played up in the barge.

Here's Shakespeare's work:

"The barge she sat in like a burnished throne, Burned on the water_ the poop was beaten gold; Purple the sails and so perfumed that The winds were lovesick with them; the oars were silver, Which to the tune of flutes kept stroke, and made The water which the beat to follow faster, As amorous of their strokes"

This is definitely no coincidence. What's more interesting is that even though it's almost a fact that Shakespeare had Plutarch open while writing this, we can actually see him doing some creative work. He added some alliteration, and for some reason we humans love to listen to a series of words that all start with the same sound.

baristaGeek··on I am a fast webpage
Bootstrap was just an example. It could Bootstrap, Materialize, W3.CSS, etc.

The point is that it's much more convenient to reuse code from a framework, because it's better to sacrifice file size for fast iteration and functionality.

← PreviousPage 6 of 9Next →