Dylan Vassallo

How we built Classyfy

2012-04-23T00:00:00-07:00

At 5 PM last Friday, roughly one hundred Bruins descended upon the Career Center to take part in the Facebook Hackathon at UCLA. With the help of friendly Facebook engineers and ridiculous amounts of food and caffeinated beverages, teams of up to four students worked through the night to build a project that would then be presented in a two-minute demo at the end of the competition. After 24 insane hours of designing, coding, debugging, and polishing, Team Out and About (Angela, Samir, Prasanth, and myself) took first place with our app Classyfy. Classyfy is a fun way to take your party photos from trashy to classy: simply log in with Facebook, choose a photo album, and watch as the red party cups in your images are detected and replaced with juice boxes, sandwiches, Red Bull cans, or even pineapples. It's not foolproof, but it's a lot of fun.

It was a great project to work on and I'm really proud of my teammates for how well we worked together in executing on our vision. In November we'll be heading up to Facebook HQ to compete against the winners from other schools, but for now I'd like to share some details about how we built Classyfy.

The Classyfy app can be thought of as two connected systems, which I'll call the "frontend" and the "backend". The frontend is a pretty standard web application, handling the Facebook authentication flow and serving rendered templates to users. The backend is where the heavy lifting of processing the images takes place. We chose to host the frontend on Heroku, which was a time saver in several big ways. Not only did we get HTTPS (required for Canvas apps) and git-based continuous deployment for free, but upon creating our app in the Facebook Developers interface, we specified that we wanted to use Heroku and in seconds were able to clone down some skeleton code that was all set for use with the Open Graph. Heroku provides this starter code in several languages including Ruby, our weapon of choice. Even better, the Ruby template is built on the Sinatra microframework, which is perfect for hackathon-sized projects, and the excellent Koala gem for accessing the Facebook API.

Instead of using the PostgreSQL database that comes with every Heroku app, we set up Redis on an Amazon EC2 instance. I had already been in love with Redis for some time, and Sinatra plus Redis is a wicked fast and flexible combination that is perfectly suited for hackathons. Adding to the awesome is Resque, a Redis-backed job queue written in Ruby that comes with a slick admin interface. Together, Sinatra, Resque, and Redis comprise Classyfy's plumbing. Users interact with the Sinatra frontend, images to be processed are pushed onto the Resque job queue, and Redis is the persistent in-memory datastore that ties everything together. Because Heroku runs in Amazon's Virginia datacenter, there was minimal latency between our frontend Heroku app running Sinatra and our Redis- and Resque-powered backend running on an EC2 instance in that same datacenter.

As for the image processing itself, we experimented with Camellia and its Ruby bindings but ended up writing a C++ program (executed by the Resque workers) that uses OpenCV and cvBlobsLib to do the red cup detection. Finding red blobs in an image is easy, but figuring out which are actually red cups isn't; for that we used several heuristics based on the shape, size, and other properties of each blob. The C++ executable outputs information about each detected red cup that is shuttled into ImageMagick for overlaying the juice boxes over the photo. Once that's finished, the modified image is uploaded to S3 and marked as ready to go in the Redis representation of the photo album. Once all the photos in the album are processed, the user is taken to the results page where they are free to share their newly classyfied images with prospective employers to their heart's content.

To: the Cloud

2012-04-12T00:00:00-07:00

I am writing this blog post from my lab on a nondescript Dell desktop computer running Ubuntu. That's not how I would typically write a blog post, if I were one for doing such a thing with any frequency. Normally I'm inseparable from my trusty three-year-old MacBook Pro, but today that machine in the care of the Geniuses at the Century City Apple Store. (Apple employee 2994584XX and whoever is in charge of AppleCare: I wish I could buy you both beers.) This is because over several months of frustrating lockups and troubleshooting I have determined that for some reason my laptop will, after some arbitrary amount of uptime, begin failing to write to its internal hard disk. It works perfectly well when booted from USB, and it can read from any drive, but after some amount of time all writes to the internal drive will begin to fail, without fail. I'm not exactly thrilled about this new behavior.

In short, a logic board replacement is definitely necessary which means that I need to do the unthinkable: part with my laptop for a period of days until it is fixed. "There are other computers in the world I can use, but none of them are mine," the voice in my head complained. "How will I finish my AI homework if I can't get in The Zone by listening to my music? Do I even stand a chance of coming up with this state machine if I can't take a Twitter break practically every single freaking minute, as I am wont to do?"

Whining aside, I was concerned about not being able to get enough work done while my computer was in the shop. But in the past few hours I have discovered that thanks to the power of ~THE CLOUD~ there is little about my setup that depends on the exact physical hardware I'm using. I'm no stranger to ~THE CLOUD~ but I was caught off guard by how thoroughly it has worked its way into my life.

So thank you, Google and Dropbox and Github and Rdio, for hosting a surprisingly large subset of my digital life in your servers in ~THE CLOUD~. Thank you, Apple, for inventing the tiny computer that lasts all day on a single charge, connects to the Internet (and ~iCLOUD~) from anywhere, and also makes phone calls. (We should really have a chat about how Time Machine is useless and FileVault is more trouble than it's worth, but overall you're pretty cool.) Thank you, Amazon, for letting me play in your Jungle Gym of Linux for pennies per hour. And thank you, hackers, for making such useful stuff (except for those of you who are off securing another round of funding for your thing that's "like Klout for Groupon!!" or whatever). I like to think I'm a bona fide hacker already, but childlike wonder still abides.

Winter books

2012-01-11T00:00:00-08:00

I love winter break. It's my chance to recover from final exams, visit family and friends, and spend time doing things I've had to put off all quarter. This year I got a lot of reading done in between interviewing for summer internships and watching Stanford CS 193P lectures. Here are the books I read, and a quick review of each of them:

Steve Jobs, Walter Isaacson: Though I think Isaacson's unfamiliarity with Silicon Valley was detrimental to the biography, his unprecedented access helps give the reader good insights into the complex person that Steve really was.

Hackers and Painters, Paul Graham: PG is a visionary and a gifted writer, and Hackers and Painters is a brilliant collection of essays about what it means to be a hacker. This is one of the best books I've read in recent memory.

Getting Real, 37signals: In this precursor to Rework, 37signals lays down their philosophies on the design, development, and business of web applications. It's a short read and well worth the time of anyone looking to successfully create and launch a new product.

Being Geek, Michael Lopp: Advertised as "the software developer's career handbook," Being Geek aims to educate us nerds about the aspects of our jobs that lie beyond the computer. There's a lot of great advice in this book, but it seems most applicable to fairly experienced engineers.

Startup School notes

2011-10-30T00:00:00-07:00

(Before I start this post, I have a confession to make: I'm probably the last person on Earth to realize that this blogging thing is hard to get into regularly. I've been doing a lot lately that's worth blogging about--research, hackathons, and the like--but you wouldn't know it by my lack of posts. I'll try to be better in the future, I promise!)

After recovering from an unpleasant case of pneumonia, last Saturday I was lucky enough to attend Startup School 2011. Produced by Y Combinator and Stanford University's BASES and held this year at Stanford's Dinkelspiel Auditorium, Startup School is a one-day series of talks by well-known entrepreneurs and investors. I found the event to be incredibly inspiring and packed with great speakers giving great advice to the audience of young hackers. All of the talks were live-streamed, and I think recordings will be available sometime soon as well; I highly recommend checking them out.

I believe it was Max Levchin who wondered aloud what happens to those compulsive note-takers after they graduate high school; I never thought I was one of them but it turns out I have a few pages' worth of notes from the event--so without further ado, here they are.

Marc Andreessen, Andreessen Horowitz
- Invented the img tag (!), started Netscape at 22
- No Startup School-alike back then; entrepreneurs didn't really know each other
- If he could go back in time, he would instrument the browser to learn how people were using it. Also would have done something like RockMelt (built-in social) as well as built-in payments.
- Building a company is like baking a cake. There are a lot of ingredients and if you're not careful you'll end up with egg on the ceiling.
- He believes in engineering at the heart of the company because ultimately it comes down to the product. And in the long run, your product is innovation.
- If the cake comes out of the oven and you forgot to add sugar, it's too late. You can't just put some sugar on top and say it's fixed.
- Hiring fast is hard.
- The "magic formula" is an engineer who can become an entrepreneur and CEO.
- Surprisingly, being CEO is a learnable skill. It's easier than learning technical skills or entrepreneurship. A motivated engineer can learn to be CEO on the fly. Zuck and Jobs are two of the best CEOs and that's their background.
James Lindenbaum, Heroku
- Building a product for developers is hard but rewarding.
- The process of making software is getting increasingly awesome-er, but deployment isn't keeping up with that.
- Wanted to abstract away deployment. Platform instead of servers.
- For Heroku, developer UX is key.
- Japanese poka-yoke: something that prevents you from making mistakes.
- After launching, they realized they'd need venture backing.
- YC is a bridge between the tech world (they're approachable as hackers) and the investing world.
- PG: "You should kill Oracle." James: "Um, okay."
- Web-based editor was seen as a toy; devs loved the new git-based deployment
- It is important to Heroku that they have code on their homepage; they want customers who get that.
- Patterns
  - App decomposition into services: EC2 allows for extremely low ping times between disparate services.
  - Polyglot platform: devs are comfortable learning new languages and using multiple languages in a project.
- Salesforce acquisition (2010): why? how?
- They get to work how they've always worked, but now they enjoy Salesforce's leverage
- "Being made fun of by Larry Ellison…is the highest honor there is."
- It's about taking your specialty and offering it as a cloud service.
- Building tools for people who are building things gives you a lot of leverage and is lucrative
- Scaling the team has been harder than scaling the platform.
Jim Goetz, Sequoia Capital
- Spectacular companies start with ambitious but unknown founders, no pedigree required.
- Search environment was crowded when Google started.
- Sequoia's 40 years of experience tell them that now is an excellent time to start a company.
- It's an order of magnitude easier to start a company now than it was a decade ago.
- 10 billion mobile connected smartphones by 2020.
- In the war between Apple and Google, devs win.
- Business model as a weapon
- "Unfair advantage" if starting a startup in Silicon Valley
- In general, VC due diligence takes a few weeks.
Ashton Kutcher
- To change the world you have to change yourself.
- Red flag in entrepreneurship: "What will I get / who will I be by solving the problem?"
- Don't jump to the effect. You have to be the cause of a new reality.
- "If you want to be the next Mark Zuckerberg, you will always be second best, because Mark Zuckerberg will always be a better Mark Zuckerberg than you."
- The story of Carl Fisher
- It's about leaving behind disruption for everyone to share, and how you can eliminate the space between people.
- agrade@gmail.com
- The biggest reason that SV/NY is attractive is that you have a certain culture of people surrounding you
- Success is really imitation plus innovation
- Startup "attitude" in LA: companies that spearhead content delivery are resistant to innovation.
Matt Mullenweg, Automattic
- There's a stage in every startup when you have more devs than users. Enlist your friends.
- Be your own most passionate user.
- "To be creative you need to be, or be sleeping with, someone who uses your software every day."
- Details matter. "There's nothing like the crucible of real usage."
- Famous 5-minute install
- "There's nothing quite as permanent as a temporary business model."
- You need a better tagline.
- Do your own support. Have support@example.com be an alias for everyone@example.com for as long as you can keep it up.
- "If you are in school, please appreciate it before you drop out."
- "I don't mind being on the Titanic, but I want to be steering it" (re: CNET)
- PG's Bayesian work influenced Akismet (blog comment anti-spam)
- Time is short.
- Never rely on a single partner for > 30% of your revenue.
Mark Pincus, Zynga
- It's hard to say when your startup has actually started.
- He tried to buy CNET (!)
- It's shocking how much of our Internet behavior is represented by one company (e.g. Facebook for social, Amazon for shopping)
- Value a single engineering day building the right product.
- Test, test, test. Build the house that you want to live in, don't build a spec house you're going to flip ASAP.
- Don't worry about your resume or exit path. Go all in.
- He's a serial entrepreneur because he wasn't able to create a sustainable company.
- You also need a plan for when everything goes right (e.g. scaling the company)
- It's hard to raise money even when you're profitable.
- Early-stage entrepreneurs sometimes only want to pursue something if it gets funded. Wrong idea.
Paul Graham, Y Combinator (office hours)
- In reading this batch of YC applications: in so many cases, the desire to start a startup precedes actually having an idea. The best startup ideas don't come from people looking for startup ideas. Look for problems, not ideas.
- It's better to do something hard if you're capable of solving hard problems.
Mark Zuckerberg, Facebook
- The coolest thing about SV is that there's no reason not to just try.
- The biggest risk is not taking one.
- 1st-gen web problems were all about scale. Now, it's hard to cache when the experience is so personalized.
- It's fundamentally valuable to have the person making product decisions also have technical knowledge about the product.
- He did hold a programming competition for hiring interns, but there was no alcohol like in The Social Network.
- A fundamental of engineering: you never want to build the same thing twice.
- Once you have a product, centralize your competencies (e.g. disparate marketing people becoming a Growth Team)
- The point of Facebook isn't the features, it's the people.
- Only sell your company if you'll still get to do what you want to do.
- Find the VCs that are aligned with you.
- Surprisingly, you can learn a lot of stuff you don't know along the way if you stay focused on the user.
- SV culture sometimes favors starting a company before you actually have anything to build.
- Huge bias: companies that succeed have passionate founders.
- People will remember what gets built on top of the FB Platform more than they remember FB itself.
- FB's future: what can you build now that all of these social connections are in place?
- SV is not the only place to be. Good for beginners though.
- He would have stayed in Boston if he were to do it all over again.
- The short-term focus of SV bothers him
- Average time spent in a job in Seattle is 2x that of SV. Non-committal culture
- Non-SV startups seem to be more focused on the long term.
- There's always an advantage to doing things differently than everyone else.
Stephen Cohen, Palantir
- "The product advice you get from VCs is worth how much you're paying for it."
- The CIA has a VC branch: In-Q-Tel
Max Levchin, Paypal
- The job of the cofounder isn't to commiserate with you. It's to provide the platform of support when you need it.
- Not having a cofounder can be detrimental. No one can hear you scream, "we're fucked." Your mom telling you that "everything will be okay" doesn't mean much in the practical sense
- Pick a cofounder that you respect. If you get a bad feeling, break up ASAP. "Whenever there's any doubt, there's no doubt."
- Problem: charismatic employee that's a net negative. Get rid of them ASAP.
- How to fire someone: "I'd like to ask you to resign."
- Classic engineering mistake: confusing hard and valuable.
- South Park underpants gnomes episode
- Plagues: fear of failure. Fear of judgment. Desire to impress someone. The only emotion you should allow yourself is passion. You can't always be thinking "How will this make me look?"
- Ignore your mistakes. The number one thing to worry about is "Am I doing what I'm good at?" Don't dwell on what you did wrong, push on with what you're good at.
Ron Conway, SV Angel
- Media and the Internet are finally converging
- Think big, Really big. Not enough people do.
- A defining entrepreneur is a product visionary who owns the mind of their customer. Decisive.
- Bootstrapping forever is great.
- Design/UX are the new intellectual property.
- He can tell a defining entrepreneur from a mile away.
Drew Houston, Dropbox
- Everything big starts small
- Everyone starts out clueless
- Get out of your comfort zone
- As a founder, you start spending more and more of your time on people stuff.
- It's easier to start out as an engineer and learn the business stuff than vice versa.
- Learn a little bit about a lot:
  - sales
  - marketing
  - finance/accounting
  - product design
  - psychology, influence, negotiation
  - organizational design
  - management and leadership
  - business strategy
  - blogs, HN, founder stories
- Take on more than you're ready to, and get used to that feeling
- The fastest way to learn about startups is to join one, NOT grad school, big companies, business school, banking, consulting, etc.
- Stack the odds in your favor. Anyone can push themselves; put yourself in a situation where you'll be pulled
- Surround yourself with kindred spirits. You're the average of your 5 closest friends.
- Get plugged into the startup world via YC
- Find the best investors
- Dropbox is looking to build the Internet filesystem
- Re: acquisition, independence is important for ubiquity.

My "Setup"

2011-07-13T00:00:00-07:00

The Setup, curated by @waferbaby, is a fascinating collection of interviews with people from all over the Internet. In one of these interviews the subject simply describes the software and hardware that they use in their personal and professional lives. I've been following site since its genesis and I am inexplicably fascinated by each and every detail shared by the site's often Internet-famous interviewees. In the spirit of introduction, I humbly present my own answers to the questions posed by The Setup. UPDATE: @waferbaby was kind enough to link to this post on the Community page of The Setup. Thanks to him for that and for maintaining such an interesting collection of interviews.

Who are you, and what do you do?

My name is Dylan and I'm a third year computer science major in the Henry Samueli School of Engineering and Applied Science at the University of California Los Angeles. I work at the UCLA Center for Embedded Networked Sensing where I develop scientific research platforms that use mobile technology as a vehicle for community involvement.

What hardware do you use?

My main machine is a 13" MacBook Pro. I ask a lot of it, and it has proven itself to be an incredibly reliable companion. My particular model is from mid 2009 and has a 2.53 GHz Core 2 Duo processor; I've bumped the RAM up to 8 GB and the hard drive to a 500 GB Seagate Momentus XT, a "hybrid hard drive" with 4 GB of flash memory. When I'm writing code I find that referring to documentation is much easier with a secondary display.

I have a black 16 GB AT&T iPhone 4, though I use it less like a cell phone and more like a beautifully designed handheld computer from the future. My school and the surrounding area are thoroughly blanketed in wifi and cellular connectivity so I'm never far away from the apps and services I rely on and enjoy. My iPhone has probably improved the efficiency of my computing life more than any other device I've ever owned.

I also own a black 16 GB wifi iPad 2, which I use mostly for consuming RSS feeds, Instapaper, and my Twitter timeline. I don't think I could bring myself to manage more than one traditional computer for personal use, but my iPad is a low-maintenance and very capable sidekick.

When I'm not using my MacBook Pro or my iOS devices they're probably nestled in something made by Incase. In my apartment I've got a 1 TB Time Capsule for Time Machine backups and fast, reliable network access. I also have a first-generation Apple TV that I use to run Ubuntu (yes, Ubuntu) and a variety of open source server software. This is facilitated by the atv-bootloader project. I had hoped this would be a low-cost and power-efficient home server solution, but it's turning out to be more trouble than it's worth (the unit runs damn hot and running Ubuntu is a hack at best).

And what software?

The all-stars of my Applications folder are Colloquy, Dropbox, Gitbox, Google Chrome, Hulu Desktop, iTunes, Kaleidoscope, Brett Terpstra's nvALT fork of Notational Velocity, Sequel Pro, TextWrangler, Things, Transmission, Twitter, VLC, and VMware Fusion. Mail.app is not on this list; in my humble opinion the Gmail web interface is the only good way to interact with email servers. I've always got at least one Terminal window open and I couldn't live without Homebrew, the "missing package manager for OS X." I'm looking forward to Lion and iCloud but I'm less enthusiastic about the inevitable incompatibilities that follow a major OS X update.

I'm running the latest iOS 5 beta and it's the bee's knees. My favorite iOS apps are Air Video, Due, Facebook, Foursquare, GoodReader, Instacast, Instapaper, Mailroom, Pastebot, Reeder, Simplenote, Things, Twitter, and Verbs. If you happen to know the developers of Instacast, Mailroom, or Verbs, please let them know that I'll be first in line if or when they release iPad versions of their apps. When I have time to kill I like to play Doodle Jump, Flight Control, Plants vs. Zombies, and Words with Friends.

What would be your dream setup?

I'm always looking to streamline my setup, but overall I'm very happy with it. Any improvements I can think of simply involve more Apple products, and that's boring and a bit scary to realize. I would however make great use of a small or medium Amazon EC2 instance even though unlimited cloud computing power isn't considered futuristic anymore.

Hello, world

2011-07-05T00:00:00-07:00

Internet, you and I been acquainted for several years now and a formal introduction on my part is long overdue. I'm Dylan, a third year computer science major at UCLA. My passion for building things began in my adolescence when my grandfather, a retired electrical engineer, would take me into his office on summer afternoons to show me how circuits worked. In my freshman year of high school I downloaded the installer for Python 2.4 and the rest, as they say, is history. I've got a long way to go before I reach neckbeard status, though, and that's why I've decided to start recording my thoughts here. It's something of a rite of passage for developers to write their own weblog platforms, and while I use Tom Preston-Werner's static site generator Jekyll and a layout forked from existing ones by John Duff and Tanner Burson, I've hacked together a system that should be pleasant to look at and just as nice for managing content.

Naturally, this begs the question: what content will I be managing with this shiny new setup? In the long term, I want this blog to be a written record of my experience as a budding software developer. This summer I have an awesome job at UCLA CENS and a bunch of ideas for side projects, and when school resumes I'll go back to absorbing knowledge at what feels like an insanely high rate. From all of this I'm bound to have a lot to share, and when I do I'll post it here.