Apple Sr Machine Learning Engineer
University of California, Davis PhD candidate, Computer Science (graphics and visualization)
Current Time 0:00
/
Duration Time -:-
Progress: NaN%

How did you get to where you are today? What is your story? What incidents and experiences shaped your career path?

Summarized By: Jeff Musk on Tue Dec 24 2019
Well, I'm not sure my career is the kind of non-typical, because I started to the software engineer after graduating from Princeton bachelor's degree in computer science. I was working as a software engineer for 10 years before I decided to pursue graduate studies at UC Davis. And I did my master's part-time before that. But at UC Davis, I was primarily focused on the visualization of data sets for scientific organization and, rendering. And while that was happening, that's when machine learning became really kind of, dominant and very popular field. There are lots of opportunities in Machine learning. I saw with my visualization projects kind of led me to the machine learning. While pursuing my PG, I switched the industry and, I start working first for company called Ancestry that comes in San Francisco. That was that. Maybe the company had lots of data but wasn't really sure how to use machine learning to actually analyze that day. I was brought initially as a visualization kind of scientist. But then I got involved in Machine Learning and I was actually one of the first people there who helped them with their relevance ranking overwritten for the search record search. Right now, that's basically using milder form a line and, after that, I kind of switched completely to machine learning. And then I left Ancestry after five years and joined Apple, the fraud detection team. And again, it was kind of similar in the sense that Apple has got lots of data, but they were growing their machine learning expertise and especially the field of fraud detection machine learning. That's where I am today.

What are the responsibilities and decisions that you handle at work? Discuss weekly hours you spend in the office, for work travel, and working from home.

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
Apple is basically working from the office. Apple doesn't like people to work from home. Sometimes you have to travel for work, but most of the time we're basically working from the offices. But the offices are located in multiple geographical regions. Well, I'm a kind of Sr. machine learning engineer. So as much of an engineer at Apple, you're expected to basically do the analytics, train the model and also deploy the model. Basically, responsible for the whole life cycle of the model. And you're also responsible for making sure the model works after you deploy. So machine learning piece which is a kind of smaller piece, but then the implementation piece production, implementation, and then you also need to monitor your model is in production.

What tools (software programs, frameworks, models, algorithms, languages) do you use at work? Do you prefer certain tools more than the others? Why?

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
I think very much depends on the project. I mean, right now, it's not a secret that everybody's using Spark and Hadoop is being used for quite some time now. It all depends on the amount of data. Sometimes you come and join companies that only use relational databases. And they have to figure out the way to transition from that kind of extracting data from whatever source you have into the platform that will help you solve your problem. It's really about figuring out what tools, what platform you need to get where you wanna go, So that's a first. We'll have to figure out what you want to do then you can make better-informed decisions. But sometimes technologies like rational databases can still be important because they address certain business needs. You can't just replace them with Hadoop because, for instance, it would not be optimized for realtime kind of transactions. So it all depends on the platform you want to use. 

What things do you like about your job? Were there any pleasant surprises?

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
Like any job there are positives and negatives, I mean it's just that particular field from detection, kind of. It's almost, reminds me of a game because you play against the adversary. So whatever you work on your machine learning solutions, even rule-based solutions, you have to know that you're working against somebody who will adopt whatever you put in place. And you can never rely on some static solutions to be good enough for a long time. And it depends again on the area of the fraud but if it's a really profitable fraud for the adversaries, they will find the way to circumvent your defenses. You always kind of compete against well-funded Berkshires. That's basically a positive challenge. 

What are the job titles of people you routinely work with inside and outside of your organization? What approaches do you find to be effective in working with them?

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
Like any big company, you work with other groups, you work with people responsible for the relationship between groups, titles, like, for instance, EPM's kind of like project managers that I work with a lot because they basically control things. They have all the schedules for other groups. Let's say you want to work on some features or some model something that will require, and very often that's the case you require something, some work from other groups. So as a machine learning engineer, you can't make other people work. You basically have to work with them, managers of those groups through their project managers and figure out how to feed that into their timeline. And that's a challenge sometimes because maybe your project requires you to deliver something as soon as possible while other groups make another priority and that's something that comes up in big companies. In smaller companies, this is less of a problem, but big companies, that's the challenge you have to accept and basically navigate. Deal with managers from other groups, project managers, a lot of project managers of your own group and your manager who has also kind of certain priorities. What things need to be worked on, things like that.

What major challenges do you face in your job and how do you handle them? Can you discuss a few accomplishments?

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
I can't talk about specifics on any projects that I've been worked on. I can just talk in general. I mean, we basically succeed with what we have. Basically, our success is determined by the metrics we set for ourselves. Now let's see when we have accomplished something fighting with fraud has to basically have certain KPIs to measure, our success. We basically have to see the KPIs progressing from year to year. But sometimes we have to revisit those KPIs. Maybe the KPI is not measuring what we want them to measure because there are always other metrics that can be can be looked at. For instance, we may think we're winning the battle which was, let's say, spam, but maybe our customers still contacting Apple care in large numbers. So maybe there's a disconnect between those two things. So it all depends on how you set a KPI. And how do you verify those improvements in those KPIs actually reflect in real life. So that's also a challenge and many of these projects there's always basically, keeping the right KPIs is a major part of the solution.

What are the recent developments in the field? How significant are these improvements over past work? What are their implications for future research & industry applications, if any?

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
Part of the new developments of the last 10 years is related to Deep Learning. Rise of Deep Learning as a very useful tool for addressing longstanding problems of object recognition, machine translation. All those things cover implications for a lot of different areas of basically software products, and plus they include fraud as well. Because in fraud, there are some projects that were ready for object detections, there are some projects were ready to checks analysis, checks memorization, for instance, like that risky. There's been a lot of progress, and it was NLP, and then we look at some areas of fraud that related to fake reviews in App Store. Or, let's say, putting some advertisements in the comment section as a review for some app. So recognizing those things automatically would really help us free up lots of people who sometimes we have to do it manually. So all those kind of developments in the planning definitely having an impact on the industry and real-world projects for sure. 

What was the hiring process like for your job? What were the roles of people who interviewed you? What kind of questions were asked?

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
It's gonna probably be irrelevant now because my I was interviewed almost four years ago and I think that's really changed, really standardized. At this point, I feel that big companies keep a very similar approach to interviewing. I guess it depends on the title, on the position. If you're interviewing for data scientists or the analysts, that could be one set of questions primarily related to the analytics background. But if you're going to interview for a machine learning engineer, it's basically an engineering interview. So it still questions just common on LeetCode. It dominates the interview process. Also asking machine learning questions. Just asking, figuring out if a person's competent pf machine learning algorithm. We also like to ask questions like basically pulls the actual real problem that we face now daily job and see how the person can address is that maybe you post a question that requires a person to ask a clarifying question. I honestly think we want to see how the candidates think their way through the kind of not a typical problem. So I want to see them analyze what we're giving them. See if they, for instance, immediately tried to propose a solution without having all their asking all the right questions. That's kind of like a red sign case. So that's pretty much it, but that it's really similar to I think a lot of big companies, Coding questions first, I think that will be disqualified if somebody cannot call or make some kind of medium or simple questions. Basically analyzing machine learning knowledge and also asking about analytics questions related to our kind of projects.

What qualities does your team look for while hiring? How does your team interview candidates?

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
You want people to be able to present their work because very often you have to sell your project. If you're starting up on some new direction and you basically need to make a proposal, you need to go through the review of that proposal saying to be able to advocate. If you're working, you need to be able to kind of justify the choices you make, and that's why actually when whenever it comes to interview for my group, for instance, they have to also present your presentation of either their research projects or the previous work that they've done for other companies. And then we just wanted to see how they can talk about their technical experience and present in a way that is understandable for non-experts.

What are different entry-level jobs and subsequent job pathways that can lead students to a position such as yours?

Based on experience at: Sr Machine Learning Engineer, Apple
Summarized By: Jeff Musk on Tue Dec 24 2019
You basically can be hired as a machine learning engineer. Apple doesn't have that many levels like some companies do. Apple really has five levels. It's very hard to get to the fifth level basically. And the six levels can distinguish. That is reserved for kind of well-known people in the industry that's not exactly expected to people to get promoted to six unless you read multiple books on the subject like that. We look at people are well educated in quantitative fields, not necessarily computer science, not necessarily machine learning. Machine Learning research they will require. Have experience in machine learning, doing machine learning research in that area. So having a degree in this area was a proven publication track record for me. But for us, because we kinda more apply, we need something quantitative, experience from web quantitative field, mostly requires people to have graduate degrees. Most people from my group have PhDs. A lot of people come with experience from other companies, either working related, like fraud detection and banks or come from companies with completely different fields as I came from the Research relevance ranking, completely unrelated to fraud detection. 

What were the responsibilities and decisions that you handled at work? Discuss weekly hours you spent in the office, for work travel, and working from home.

Based on experience at: Principal Data Scientist and Group Lead, Ancestry
Summarized By: Jeff Musk on Tue Dec 24 2019
At Ancestry, I was just started out as a kind of visualization researcher and I basically progressed to kind of a group leader of aging relevance ranking search Data Science group

What major challenges did you face in your job and how did you handle them? Can you discuss a few accomplishments?

Based on experience at: Principal Data Scientist and Group Lead, Ancestry
Summarized By: Jeff Musk on Tue Dec 24 2019
When I joined it was a company that had very little machine learning experience. There were a few people who actually we're working machine learning projects. Even there, the infrastructure was not ready to be used by machine learning experts. So we had to wear multiple hats. We had to be a part of a data engineer, required to be kind of project manager figuring out how to move data in addition to the infrastructure, making sure we have the right infrastructure and then working with search engineers and picking up what kind of search strategies we should use for that particular data and setting it framework for comparing with the existing search engine convention and using primarily can rule-based method for ranking. That doesn't work well because, basically it requires people to constantly manually update weights, which is exactly what you don't want to do with you when you can replace it with the machine learning all-day.

What was the hiring process like for your job? What were the roles of people who interviewed you? What kind of questions were asked?

Based on experience at: Principal Data Scientist and Group Lead, Ancestry
Summarized By: Jeff Musk on Tue Dec 24 2019
Actually, they basically hire high-level managers first to kinda grow the team that would be working on big analysis and hopefully search. So basically, I was one of the few people who was brought in because of my interest in visualization, because they have lots of data that they want to visualize. And some other people were hired because of their experience in search and other fields. So had a cross-disciplinary team. It was a kinda real machine learning researcher also joining the group. So basically, people who brought along various experiences also was brought in because of my self engineering experience. 

How did the program prepare you for your career? Think about faculty, resources, alumni, exposure & networking. What were the best parts?

Based on experience at: PhD candidate, Computer Science (graphics and visualization), University of California, Davis
Summarized By: Jeff Musk on Tue Dec 24 2019
It was a very well known program in visualization. I think Davis and the University of Utah are kind of top schools on visualization. So basically I joined them coz I already completed my Master's in visualization so I could have progressed UC Davis for a PhD program continuing in that ring, but kind of switching the different data sets. I was working at UC Santa Cruz from my Masters. I was working in a kind of computational fluid dynamics specialization while in Davis I started working on things like climate data sets and things like that. But very quickly I realized, the base I was visualizing data produced by scientists from other fields. It was a secondary thing. It was more like producing tools for scientists to use on, especially when I was switching more to work with machine learning scientists and work on visualizing their data sets, I realized my interest actually close that lying to being in machine learning itself. So it kinda helped me to realize what I really wanted to work on. And that's why I kind of slowly transitioned into the industry. And my PhD is on hold right now. I can get it, but it's on hold. I'm still trying to figure out how to merge my field of study with my actual work in the industry. 

How did the program prepare you for your career? Think about faculty, resources, alumni, exposure & networking. What were the best parts?

Based on experience at: MS, Computer Science, University of California, Santa Cruz
Summarized By: Jeff Musk on Tue Dec 24 2019
I basically went to my undergrad in Santa Cruz and the grads. It had a good computer graphics visualization program I was interested in that when I was doing my bachelor's as well as when I was doing my master's. So that was the kind of interesting thing for me and that's what I was working on there.