Lu Interview Results: Fill & Download for Free

I’ve gone through a few of the interviews (Microsoft, Google, Uber, etc.), and I can tell you they can be quite different. The Uber interview was probably the most machine learning focused, and I was asked questions about my background in ML, various algorithms such as SVM and K-means, along with some basic CV knowledge.The Google one was actually for a SWE intern position, but once I passed the technical interviews I got some more project specific interviews. Among these project specific interviews were two machine learning projects. For these two interviews I was just asked about my background in ML and some of the past projects and research I had done. I ended up choosing the ML project on the Search team (the project involved ranking and intent prediction).The Microsoft interview was also more technical focused, but I was asked a few questions about general ML concepts. At the internship itself I worked on a lot of NLP, such as language modeling and dialogue systems.The experiences I had doing ML as an intern at Microsoft and Google were really interesting. Before I had only done research in ML at school, so I was really focused on understanding all the theory well and being able to come up with some novel ways to improve a model’s performance. However, in industry I was surprised to see that the focus was pretty much just on the code itself, and understanding how to apply and tune models. I probably spent 70% of my time doing data processing, model tuning, and results analysis. Even the full-time ML engineers spent most of their time on these three things.From my internship I actually realized that industry ML (particularly ML engineering) is more about knowing some general model architectures, when to apply them, and how to interpret results in the context of a problem. Also, a ton of data analysis and processing, along with model tuning.When I was learning ML for industry, it took me a long time to read through all the documentation and online tutorials. It was also incredibly difficult to filter out the relevant information to my work. Because of this and my internship experiences, I created my startup (AdaptiLab), so that future aspiring ML interns/engineers can learn the skills directly related to industry work, in a much more streamlined and easy to understand process than what I (and many others) had to go through.

Developing a resistance to rejection is impossible.Ever heard of hedonic adaptation? That we will inevitably return to a baseline happiness level, regardless of life circumstance (with some exceptions)? Fear of rejection is the same way. Being rejected 1,500 times is not like learning to ride a bicycle. It's more like stretching out a rubber band: you can pull it and wear it down and increase its elasticity, but eventually, it always shrinks back to its normal size.There are, however, tactics you can employ to stretch the rubber band faster, and for longer periods of time.1. Develop aphorisms that encourage you to act, and work on getting your brain to truly believe them.You know sometimes, in a late-night conversation with a friend, there comes a point where both of you stumble upon a placid, short, completely unoriginal idea that, as a result of the build-up of your conversation, becomes a universe-maker? Develop the ability to come to these conclusions by yourself, and through faster channels. I have a couple of sayings, that, by thinking about frequently, I've come to believe more and more.Nothing is ever such a big deal.There are literally no consequences to failure.If I'm not having fun, I'm doing something wrong.There are infinite possibilities.These days, I find myself thinking about going into a situation with a "200% mentality": shoot for completely winning over the person two-fold, because even if I don't hit my target, heck, I'll be at 100%.These adages might not hit you perfectly; you're going to have to go through an internal process to discover and develop what works. Stay relentlessly optimistic and be irrationally confident. If you're stuck, just say to yourself, out loud, "I have massive, Atlas-sized balls."2. Place yourself in a state of natural gregariousness.I've learned in psychology (literally from the professors who've come up with these theories) that environment often plays a stronger role in determining behavior than personality. We act and react based on what's happening around us. If every male between 17th and 18th avenue in Manhattan was approaching a beautiful woman, you'd be more inclined to as well. (Google "fundamental attribution error" for slightly more useful information.) The takeaway? When you're about to face life, jolt yourself with little bursts of confidence.Play your favorite song (Lose Yourself, by Eminem, seems to be a perennial contender in this department)Talk to someone you're already comfortable with (for me, it's calling up a good friend)Watch an inspiring video (as a swimmer, it's the Beijing Olympics Phelps-Cavic miracle finish)Brainstorm a list of quick personal activities that place you in a buoyant mood, and before a big potential "fear of failure" moment, actually try them out. If they work, tack it in the "Approved" pile. If it doesn't, scrap it.An additional tip: the facial feedback hypothesis theorizes that the physiological act of smiling -- your muscles turning your lips upwards and tightening the skin around your eyes (that's a Duchenne smile) -- will affect your mental processes, and make you happier.So smile, find an excuse to laugh, mime your future actions, exercise those mirror neurons, dammit, and do it over and over again. When a friend asks you how you feel, respond, "I feel like a champion"; when someone at a party asks you what you majored in, tell them, "I majored in unafraid." Walk into a situation already smirking, and the rest, you'll find, will flow much easier.3. Find positive people who will encourage you.I think this is the most important point by far. A lone cyclist almost never does as well than if he races with a team (even removing wind drag effects), and there's a reason why superheroes have sidekicks. Find a wingman or wingwoman who is caring, outgoing, and resilient, and who you like enough that you want to see him or her succeed as well.The great thing about surrounding yourself with positive people is that it's both a local and global effect: you'll develop better cadences in life knowing people else has your back, and you'll be more joyful in the short-term because of their presence. Not only will they provide a home base to come back to after a failure (for sympathy) or success (for affirmation), but more importantly, they will push you to do things you normally wouldn't have the courage to do. Namely, they'll help you get over your fear of failure in whatever realm it is.Right now, it's 3 p.m. in Manila, and looking over the city on the 23rd floor of my office building, there's large-and-in-charge billboard that says, "Lucky Me!" That's a positive environmental cue right there, and what it's telling me, in no uncertain terms, is that the world is my oyster. Yours too. So get out there and do it!

CANDIDATE: Wow, that’s pretty serious. Let’s figure this out. Is this error expected? Do we historically have 10% error on this page?INTERVIEWER: No, this is a new thing.CANDIDATE: Okay. What is the usual error rate on here? And when did this start happening?INTERVIEWER: The error rate on this page is less than 0.01%. This started happening last month.CANDIDATE: Hmm, okay. Is this a gradual increase or a sudden spike?INTERVIEWER: It’s a sudden spike.CANDIDATE: Okay. Errors are not what we want or expect on a page like this. Seems like a real problem. Let’s mark down the date this happened in. We should remember it because it’ll be important later on. Let’s figure out what caused this problem. I would do a quick skim through my emails for things around that day to see if we’ve pushed out any changes that could have caused this. This is just a shortcut that won’t take too much time.INTERVIEWER: You found a few changes being pushed out around that day, but nothing incriminating.CANDIDATE: Okay. It’s possible some of those changes inadvertently caused it, but let’s keep going. I want to be absolutely certain nothing else is out of the ordinary before I dive deeper, so I am going to pull the funnel here. It’s likely something like: DAU on Facebook, views of all pages, views of this page, error rate. A caveat here is that I assume even an error on this page is still a view.INTERVIEWER: Everything looks normal, except the error rate.CANDIDATE: Alright, let’s move on then. Let’s break down this by source… by device first – desktop, iOS, Android, and Kindle.INTERVIEWER: This error seems to be happening on everything.CANDIDATE: That possibility is excluded. Let’s see if this is happening on all browsers.INTERVIEWER: It seems to be fine across browsers.CANDIDATE: Okay, we got another source we can breakdown by. Can we break this down by direct (people coming directly here), external search (searching via Google), Facebook search (internal searching), and referral (paged linked elsewhere).INTERVIEWER: Turns out the error is happening exclusively from referral.CANDIDATE: Okay, let’s break that down further. I am imagining there might be multiple sources of this, such as buttons, or links internally and externally. Can we break it down further by that?INTERVIEWER: Seems like the error is exclusively from a specific button! Turns out one of the changes you saw in the email earlier caused this. Good job!