HNHacker News
TopNewBestAskShowJobs

cjalmeida

739 karma · joined February 23, 2015

Software developer. Find me at cjalmeida(at)gmail.com
submissionscomments
cjalmeida··on Magic Leap Ships First Set of Devices Under Tight Security Constraints
That's always the case nowadays, huh? I think the days of startups CEO earning engineer level salaries and flying coach are long gone - at least for VC funded companies.
cjalmeida··on Wapp – a single-file web framework by the creator of SQLite
IIRC Falcon only does REST, it has no template rendering.

(To me, that's a plus)

cjalmeida··on Wapp – a single-file web framework by the creator of SQLite
YAML can be pretty confusing at times.
cjalmeida··on Trump Hits China With Stiff Trade Measures
It doesn't goes both ways. China is not a democracy and Mr. Xi is increasingly more powerfull. Trump is way below 50% ratings and Republicans are behind in the generic polls.

This trade war is not about who gets hurt more. It's about who can sustain being hurt for longer time. There's a major election in 7 months, so I'm betting on China.

cjalmeida··on Trump Hits China With Stiff Trade Measures
There are argument for protectionism, just not for a highly developed country such as USA.

But I'm not going over them for a single reason. This act will fail "bigly". China is the 3rd largest export market for USA. It imports a lot of grain and cars and can retaliate on politically sensitive goods.

Midterms are coming and China may use it to restrict tourism to Florida, put tarifs on soybeans from midwest, cars from Michigan. Companies from those states will likely suffer more in short term than benefit from the protectionism - it takes a while to build up capacity and conquer markerts. Meanwhile people will put the blame on Republicans as companies start the lay offs.

cjalmeida··on An Italian Region Where Co-Ops Produce a Third of GDP (2016)
I see. But that also mean the original investors - the ones that put the initial money - are getting sub par returns. Or government land subsidies play an important role here assuming that private developers have no access to it.

I don't see the lower returns as particularly damaging - there has been too much speculation in real state in recent decades. But that might not be the case once (if) returns get lower.

cjalmeida··on Britain’s White-Collar Cops Are Getting Too Good at Their Job
Your grandma was right.
cjalmeida··on An Italian Region Where Co-Ops Produce a Third of GDP (2016)
I'm really curious why co-op building companies provide cheaper housing compared to regular private companies.

Public sector inefficiency is well studied. Co-ops efficiency however not so much, at least for me.

cjalmeida··on Britain’s White-Collar Cops Are Getting Too Good at Their Job
Visit any 3rd world country to know what being a real prole means.

I have a hard time being sympathetic to folks that were born in a stable an generally safe country, with access to decent (if not prime) education. And all they care during their school years is girls and pot/meth.

Than they blame their unemployment on piss poor migrants that can't even speak the language, or asians that studied their asses off with the little they could make.

There are good people becoming unemployed due to tech advances. The solution is having them learn new trades - not expelling migrants so they can clean bathrooms for $15/h

cjalmeida··on Britain’s White-Collar Cops Are Getting Too Good at Their Job
Most of those low paying jobs are taken by immigrants, who are happy to do them given the alternative. Natives, supposedly with better education and jobs, reap the benefits. It's a win-win situation.

You don't need to be posh to benefit from migrants cheaply building your home, cleaning your bathroom, or taking care of the children and the elders.

cjalmeida··on Britain’s White-Collar Cops Are Getting Too Good at Their Job
>First, dumping cheap labor and goods into a country has a very real impact on poor and working class people

Yup. It keeps inflation in check, and allows them to buy more stuff with their salary, including investing into improving their education and don't actually having to work on low paying jobs.

Sometimes I wonder if David Ricardo was from Mars.

cjalmeida··on Operation Serenata de Amor – An AI project to analyze public spending in Brazil
You're wrong. In Brazil corporate income is taxed at 34,5% (base 15%, plus additional 10% for medium to big sized companies, plus 9,5% for social security)

That amount is on par or above OCDE counties, and much higher than most developing nations.

What we don't have is dividend tax. Unlike corporate tax that one must pay every quarter, you only pay dividend tax if you give money back to shareholders, so reinvestments are not taxed.

Usually countries balance income and dividend taxes so the government has a stable income but doesn't penalize investments as much. In Brazil, we opted for screwing investors.

Not to mention that companies pay an enormous amount of money in combined sales taxes. A manufacturing company usually pays 17% for state sales tax, another 9,35% federal social security, and sometimes an extra 5% to 15% for "luxury good" (like shampoo).

So, again, you're completely wrong and companies (read YOU) pay a hell lot of taxes in Brazil.

Source: I worked in investment analysis in another life.

cjalmeida··on Scotland’s floating wind farm is showing how powerful offshore wind can be
The denominator is based on standard test conditions. Check the link for details.

http://sinovoltaics.com/learning-center/quality/standard-tes...

Those conditions reflect not the theoretical maximum but an expected output in very good actual conditions.

This is relevant because it relates to the needed transmission capacity one need installed. For instance, at 30% capacity factor, you're "wasting 70%" of transmission on average.

It's a good metric to compare between sites of a same energy source but it's not very useful between sources. For that, the better metric is Levelized Cost of Energy (LCOE) in $/MWh. Usually present with or without subsidies.

cjalmeida··on Scotland’s floating wind farm is showing how powerful offshore wind can be
Floating offshore wind has it's own manufacturing dynamics. For instance, you can build the whole structure in a shipyard and toll it, unlike onshore.

Floating towers are definitely a revolution in wind farm building.

cjalmeida··on Canada to Scrap IBM Payroll Plan Gone Awry Costing $1B
This problem was likely better served by many smaller systems that better fit the specific needs.

Government is way too complex to try to cram all rules into a single monolith like Oracle's PeopleSoft.

cjalmeida··on 23,000 HTTPS certificates axed after CEO e-mails private keys
Granted, but in corporate IT, where stakes are higher, those concession should be minimal. Like, setting up a certificate with only a CSR is server setup 101.
cjalmeida··on Machine Learning Crash Course
OTOH, automating feature engineering is a thing. There are papers on using unsupervised methods to do that.

The 1st place in Kaggle's Porto Seguro competition trained an Autoencoder on raw data to extract features.

cjalmeida··on How not to run a CA
Worse yet, the tool seems to run JS on the client side and the page includes JS code from 5 different sources, including ad companies.

Any of them can capture the generated key.

cjalmeida··on 23,000 HTTPS certificates axed after CEO e-mails private keys
You need private key for the certificate management protocol. Maybe the had some arrangements for automatic revocation back when Symantec was running the CA.

Anyway, you are compromising security over convenience. A trade-off IT should never accept.

cjalmeida··on 23,000 HTTPS certificates axed after CEO e-mails private keys
They said in the article they kept the private key to simplify the revokation process. Thus, very likely they also generated the private key.

It's a bad practice promoted by issuers that want to sell a "managed" or "value added" service corporate IT is so fond of. The lesson here is that there's a lot of IT people that don't grasp the basics of key management.

cjalmeida··on Ask HN: What do you look for in a remote developer?
Seconded. I like to take a break and go to nearby park with a coffeeshop just to see other people. I don't miss office interaction at all and I have good off hours social life.

And discussing tech stuff at HN is better than doing it with "average" peers at work anyway.

cjalmeida··on Ask HN: What would you pay $10,000 for right now?
>I would pay $10,000 for an API that classifies an article into a topic such as politics, tech, etc.

That's fairly doable at decent accuracy with using current NLP techniques. At that budget I guess you can find a smart freelance dev to do the job (wink ;)

cjalmeida··on Propel – Machine learning for Javascript
There's no python vs python3 decision to make. There's no (sane) reason to pick Python 2 in 2018 for new projects.

Conda installer comes with Python prepackaged. So it's really a single exe and one CMD to get a Tensorflow env setup. And, of course, you have access to numpy, pandas, jupyter, PIL ...

cjalmeida··on Propel – Machine learning for Javascript
Python has a larger ecosystem for data intensive application and pratictoners. As a language it's very easy to learn for JavaScript devs.

The only "killer" feature JS may bring to the table would be client side, browser based inference that leverages GPU. But we're eons away from that.

cjalmeida··on Dropbox S-1
Oh, that's largely a myth. These seller do have some influence, but it's much less important than what they want you to perceived. Fashion tastes comes from a lot of sources and some is part of your identity.

That's part of what makes this industry hard.

cjalmeida··on Dropbox S-1
You could initially present a user some images to assess it's preferences and use some recommender system (collaborative filtering), a la Netflix.

That could give you an starting point. But I believe the main issue is that fashion products have, by definition, short shelf life, so you can't run the algos on SKU data. Then you can use deep learning on product images + user categorical data to try to predict preferences, maybe simple binary classification?

I guess using images as input should give better features than textual description.

cjalmeida··on Dropbox S-1
This seems huge considering how popular ETF are these days.
cjalmeida··on Dropbox S-1
I've worked in fashion before and I'm a machine learning pratictoner now.

I never tried myself but it seems quite feasible to build a decent profile of someones taste in fashion from a bit of data.

Fast fashion gave us the logistics (no more 9mo from concept to store). But Zara and friends still supply only the major trends. We're still missing for someone to reliably market the "long tail".

cjalmeida··on A Global Optimization Algorithm Worth Using
Dlib saved me a few times and I have a lot to thank David.

However I just wish it's Python binding were published as binary wheels. Dlib is very large and the compilation times suck when installing from pip.

In fact, I might as well contact him and help setup this!

cjalmeida··on A promenade of PyTorch
IMO, shuffle is something they did really fine. Unlike PyTorch datasets, TF allows streaming unbounded data. For something like this work with shuffle, it must cache some data before passing it down the pipeline. You specify how much in the argument.

This may not seem useful this conventional training, where you usually work with a fixed amount of samples you know beforehand. But there may be cases where this is not true (for instance, in some special cases of augmentation) - the streaming part is useful but then you must use this caching trick.

But I agree API naming is not stellar, or at least should come with better documentation.

← PreviousPage 6 of 10Next →