Panel: Career Advice for Data Scientists | RStudio (2020)
[CHIME RINGING] JEN HECHT: Hi, everyone. I'm Jen Hecht and I'm the VP of People Operations at RStudio. And I'm thrilled to be on this stage today with some folks who are going to help us talk about building a career in data science, in particular around the use of R, but not exclusively R.
So let me just introduce the panelists that we have today. Sitting right next to me, we have Sydeaka Watson. She is the founder at Korelasi Data Insights and also senior data scientist at Elicit Insights. We have David Keyes, who's a founder of R for the Rest of Us. And we also have Gabi de Queiroz, senior engineering and data science manager at IBM, and also founder at AI Inclusive and R-Ladies. So, welcome.
[APPLAUSE] Yeah. Very cool. So I thought we'd get started by asking each of you to talk a little bit about your own experiences sort of growing your career in data science. We've heard, I think, a few panelists today, or speakers today, talk about what is it to be a data scientist.
I just heard a talk about salary structures. And I think it's a topic that's relevant for a lot of people in different ways. I'm interested in hearing a little bit about how did each of you experience milestones in your data science career? Or perhaps if you're a team manager or somebody working with people that are learning as data science, what are the stages of this and how have you moved through them? GABI DE QUEIROZ: Yeah.
[Pause] GABI DE QUEIROZ: On. On. JEN HECHT: All right, we have a-- GABI DE QUEIROZ: Hello, hello? Hello? JEN HECHT: Can you hear Gabi? GABI DE QUEIROZ: Hello? OK.
Try this one. Hello? No. OK. All right. All right.
This one is working. Cool. So let me just summarize kind of like my career path, because it's interesting and different from all of us here.
So I was in academia for about five years doing research with air pollution data. So then I moved to the industry. So I had this transition point where I'm like, I want to try something else. I want to go to work for a company.
So then I became a data scientist. So I was a statistician before doing research and then I became a data scientist. Also, I moved to San Francisco. So statisticians in San Francisco, they all became data scientists. So I was first working for a startup and then doing some kind of like consulting. And then I moved to a different startup, and then became a lead data scientist, and then moved again, senior data scientist, moved again, became a developer advocate, and then I got to a point, an opportunity to become a manager.
So it's kind of like I don't know exactly what are the-- you wait two years and then you did this. It's kind of like, it's when I felt ready. And I think becoming a manager was a big jump from all the other transitions that I had to go through. So I felt like, I think I'm ready to become a manager. Let's see how it goes. JEN HECHT: Yeah.
OK, great. DAVID KEYES: [INAUDIBLE]. SYDEAKA WATSON: Is it working? Nope. DAVID KEYES: Try [INAUDIBLE]. SYDEAKA WATSON: Hello, hello.
OK, all right. Hi. So my path into data science definitely has not been linear, particularly because my motivations, my just thoughts about who I was and what I wanted to become and the types of things I wanted to do, they changed. So for example, when I was in high school, I thought, I love math. The people that I see around me in high school are my math teachers. They love math, and they became math teachers, and so I'll be a math teacher.
And then at some point, I discovered that there were some other things beyond education that you could do with math, and heard about being a math professor, and then found out about biostatistics after I became interested in that type of career. So I ended up becoming a biostatistics researcher. And then as other people have drank that same data science Kool-Aid, heard about all of the cool projects and the great salary structures associated with data science, and decided that that was something that I wanted to do. So some of it was just exposure, where people told me about other different types of paths that I might not have thought about, and some of it was just also just restlessness, change, and interest.
Maybe I was really happy with my career as a biostatistician for the most part. I didn't have a problem with my co-workers or the work. But it was just, I kind of want to see what else is out there, and I keep hearing about this other thing, so let me just try that. JEN HECHT: That's great. DAVID KEYES: I think this one-- can you hear me? Yeah. Great.
So first of all, I should say, I think my current work as well as my path is pretty different from the other folks on this stage. Unlike them, I work for myself. My business, as Jen mentioned, is R for the Rest of Us. And I primarily focus on training folks to use R. And I particularly focus--
I think of it as kind of like people who are scared about R, but want to learn it. That's kind of my niche. And how I got into it, it's very much by accident, which I know is a story that resonates with a lot of folks here.
I'm not very quantitatively trained. I have a PhD in anthropology. My dissertation was entirely qualitative. Worked on some mixed methods projects, kind of butted-- or not butted up against, but came in contact with some folks who were using R.
But never actually used R in grad school. And it was actually only when I started doing some consulting work where I was doing what's called program evaluation work, looking-- it was kind of applied social science research, looking at the impact of programs run by non-profits, that type of thing. And the organization I was working for allowed me to use whatever software I wanted to. So I taught myself R. And basically since then, I realized that there was a need for other people in that industry as well as others to learn R. And so I have kind of made that my thing.
But now I forgot what your original question was. So if you want to-- JEN HECHT: Well, we're free to divert from my original question. It's been interesting to hear all the stories.
I think one theme is that it's non-linear. It's not like there's a set of predefined steps or ladder or you got to do this before you move on to the next thing. You hit on something that when you were talking about having to teach yourself and not feeling like you were already part of a community that was very quantitative in nature.
And I think that's another question I had for all of you. How did you build that kind of community, whether you were in industry, when you were learning, or you're coming out of academia, you kind of go out and find it for yourself? I know all of you, in your own ways, have been involved in trying to bring other people through that journey. But could you talk a little bit about how you started to find that for yourself and how that evolved? GABI DE QUEIROZ: Yeah, so I did not have this community before back in Brazil. So when I moved over here and created R-Ladies, that was one of the intentions. It's like, I don't have a community, so I'm going to create it. So that's how it worked for me.
SYDEAKA WATSON: So I could answer this question in two different ways, according to the period of my life that it relates to. So in the earlier periods, of course, where I'm new to the different types of mathematical concepts and different statistical-- I guess I didn't mention this before. So my background is in math and statistics. So I have a bachelor's in math, a master's in math, a master's in statistics, and a PhD in statistics, and no student loans. Yay. [LAUGHTER] And so all of those different types of educational experiences put me into different types of environments, where I've got a particular mix of students, a particular set of faculty, who are trying to, together, help me to learn what I need to learn from whatever that perspective is.
And so in some sense, that community was curated for me just because of where I was in school. But one thing that I've noticed as a professional is that you don't always have the benefit of having those readily made communities. Sometimes you just have to either create your own community, as our esteemed panelist did for R-Ladies. I'm also an R-Ladies Dallas chapter organizer as well.
So definitely understand the value of those. But sometimes you have to be just very comfortable being a community of size one, and especially in places where maybe you're the only one who cares about data, or you're the only one who cares about R or whatever this new model or tool is or whatever, where you're either internally motivated, and you're just reading and going online, looking at blogs or different types of user groups, or you're also just going out, finding meetup groups, finding different types of organizations, just to connect with people who have similar interests to you. So I think it helps to develop yourself internally and also externally. DAVID KEYES: And for me, I think the answer is actually pretty straightforward. Online communities, and particularly Twitter, have been really fundamental in terms of my learning of R-- I mean, I can say without reservation that I've learned more about R from Twitter than any other single source-- but also, finding a community and building connections with folks. I've met so many folks here who I have known through Twitter.
And I've only used R for three years, only really kind of got more involved with this community in the last couple years. The major benefit, if you're not already aware, of the Twitter community and the R community in general is that it's extremely welcoming. And so I know I felt coming into it, like, I'm an anthropologist. I don't have much quantitative background.
Am I going to be accepted? And I've never not felt accepted. I think because there's such a diversity of the things that people do with R and a number of other reasons, including, strongly, the work of R-Ladies, have made the R community extremely welcoming. And so I think putting yourself out there in that community, even when you feel like, you know, who am I to try and have anything to say about the world of R when there's so many people doing so many amazing things, it is such a welcoming community and it is really possible to find a place in it. JEN HECHT: That's great.
I'm smiling, because I've had so many conversations with people here. And I think almost everyone feels that way. One of the great things about coming to the conference is you get to hear amazing presentations from people doing cool things. But I think the dark side is people may be wandering around going, oh, I'm one that doesn't fit. I have one other prepared question.
And before I ask it, I want to just let people know, you can ask questions for our panelists on Sli.do I think this is probably pretty routine for folks now. We're in track one and the code is Hexagon. So this is a pretty general question. But let's make a broad assumption that a lot of people in the room are maybe in the earlier stages of their data science career.
If you had, like, a few minutes to spend with them and give them advice about how to grow that in the best way possible, what advice would you share with people? SYDEAKA WATSON: So I can start. So my background, I mentioned that I actually went to school for statistics, and actually went into a biostats career, and that was before I even knew what data science was. So when I learned what it was, I realized, especially as I started looking through some of the different job opportunities, that they would ask for this long list of skills. And it was very intimidating. And we need a person who knows this and they know this and know this.
And it was just, how can I get that expertise so that I can just get my foot in the door into this first role? And so that would be one thing that I would advise you to think about is to just not be overwhelmed. Just think about some of the key principles, I guess, that I would say are fundamental to data science. And I could list it out as a bulleted list if you want to write them down. I would say that there are three key areas of data science that you'd want to have some proficiency in.
One would be, there's obviously the mathematical component. So there's where your machine learning, or maybe even your deep learning, your clustering algorithms, all those kinds of different algorithms, even logistic regression, or ordinary least squares regression, basic data analysis-- those types of areas are very important to get some proficiency in. The next would be programming. So obviously having proficiency in something like R or Python. Some type of general use language, such as one of those, is ideal.
Not just understanding specifics of how to use this package in that module, but knowing how to program, how to be a programmer, how to actually do the loops and if-then statements and actually put together some piece of code that accomplishes some task. So that would be the second thing. And the third, I would say, would be the big data technologies. So some of you might be familiar with SQL or Hadoop, Spark, Hive, things in that ecosystem.
So having some familiarity at least on the surface level in each of those three areas, and then developing those more and more over time as you get into whichever specific role and you learn the specific needs of that role, I think that would be a good place for you to start. JEN HECHT: Thank you. GABI DE QUEIROZ: Yeah.
I agree with the job descriptions and all that. But I think if you are beginner, even, like, big data, some companies, they are not there yet and they are still on the Excel spreadsheets, CSV files. So if big data, it's like a big thing, like, you are not there yet, don't worry, because I still think there are some several companies that are still doing small data, working with small data.
But yeah, the job description is, like, don't even bother to learn all the technologies, because there is no way-- like, I have no idea how people know all that. That is the Unicorn, right? So the companies, they want this Unicorn, and you as a data scientist, you don't have to know everything, because if they are looking for the Unicorn and you are the Unicorn, you're probably not going to work for that company. You are going to be working for a different company, either. So if you are trying to get into the data science, I would say apply for jobs. You are going to feel that are you not ready and that's a feeling that you are going to have forever. You are never ready.
So just apply, apply, apply, apply, and try to get a sense of how the interview processes are, because they vary so much. Like, some companies are going to be asking questions around probability. Other companies are going to ask you questions more general, like, just to get your thinking, how you think about the problems. So I just would say, if you are looking for a job, just apply. Don't wait, because you will never be ready. DAVID KEYES: So my answer will be very much tailored to my work, which is basically as a consultant.
I think I would offer two pieces of advice. One is to try and find a niche if you are thinking about trying to do some consulting. One of the major advantages of R is it's multifaceted, right? You can do so many different things with R. And as an R user, you know that and you get very excited.
I think I've found, as a consultant working with folks, it can be overwhelming to come in-- because I do training and then I also sometimes actually will do some work for folks. I'll get really excited and I'll say, well, we can do this in R and that in R and this in these 10 other things, and that can be overwhelming for folks. And so I think really kind of like trying to focus on what you can particularly offer to a client if you are thinking about doing some consulting. So for example, I focus, when I'm not training, I mostly do data visualization, because that's an area that I'm pretty strong in.
So having that niche, I think, is important. The-- well, I totally blanked on what my other thing was. I will come back to it if I remember. JEN HECHT: OK. Cool.
I'll just add that in my experience working in HR, most job descriptions are terrible. [LAUGHS] Honestly. Like, especially in data science, half the things on there, there's no human that could possibly do it. So you shouldn't let that get in your head. I'll just add that.
OK, so with your permission, panel, I'm going to go to some questions from our folks out here. The first one has 30 people asking this question. I'm trying to decide whether to maintain a technical role or try to shift into management and decision-making. Any tips on what to do if I want to do both? SYDEAKA WATSON: Yeah. That's a struggle that I go through. And before becoming a manager was-- the thing that I was most afraid of is, like, am I going to be in this sea of bureaucracy and then I'll lose all my technical skills that I worked so hard for several years? I was not sure.
So I talked to a few folks in the company, especially the company that I work for, asking them, how was your path? And then how do you feel about keeping up with the technical piece as well? So in large corporations, in particular, you don't have to go one way or the other. You can kind of go in the middle. And you can also, like, always go back and forth. So you can become a manager and go for the management track, but you can also come back to the technical track for a few years, and then come back again to the management track.
So that's the beauty of some companies where you can go back and forth, so you keep developing. The other thing is try to get some time for you. Like, let's say on Fridays, I try to schedule no meetings. And then I go to our corner in the room in the office, and then I put my headphone on, and then I just do some coding or code review. So that's my strategy to keep me happy, because there is a piece of me that I miss if I'm not doing technical work. JEN HECHT: Yeah.
Makes sense. SYDEAKA WATSON: I've had that same fear, actually. So I was actually kind of interested to hear what other people had to say about this, because I've been invited to go into management and into director-level roles, but I like having my hands on the data.
I like having my finger on the pulse of all the different types of cool tools and strategies that are out there. And so I don't want to lose that. So part of me is nervous. So I haven't made that jump. But I would say that was something that-- I really like those tips-- doing things that help you to continue to work those muscles. So I know some people who actively had to go on HackerRank every day and do some coding exercises in the morning or actually participate in code reviews and so on.
So I would imagine that would be a really good strategy. Yeah. GABI DE QUEIROZ: Just one thing that I was thinking now, so if there is anything that I need to measure-- let's say I need to measure something on my team and then I'm like, yes, I have some time for me to create our shiny where I'm going to measure something.
So I try to translate tasks that I need to do in a coding exercise, so I keep up with my coding. Yeah. We can talk. Like, it's a big jump. It's a big jump for sure. It's so different.
But also, you are developing other pieces of your brain, your skills, that is going to help you also in the future as a coder, programmer. So another example that I was thinking is one day I had to write something. And then it came.
Like, things were coming through my mind. And I'm like, OK, I didn't forget those things. And then I became a better searcher, or I don't know how to say that. But I can now search even better, because there is some piece of my brain that got developed a little bit better, so I know exactly how to look for answers, where before I would struggle much more.
So I don't know. Yeah, it's something interesting, I think. And if you have the ability to, you know, I want to become a manager, I'm going to give, like, a time-- I'm going to timebox this for two years.
I'm going to try it out for two years. It's like, if I don't like it, can I go back? JEN HECHT: Yeah. That's great. And David, from what you were saying before, you have your hands in doing direct work as a consultant and data visualization, and you're teaching.
So you've been able to combine those a bit as well. DAVID KEYES: And if you work for yourself, you get to do both. JEN HECHT: Yeah.
DAVID KEYES: Better or worse. JEN HECHT: Yeah, that's great. OK, another very popular question up here is, can you talk about salary negotiations? Any tips? SYDEAKA WATSON: OK, I'll start. [LAUGHS] All right, so I'll start with a story, something that actually happened.
So I was a contractor working at a company, along with a friend, a male colleague, who was also a contractor. And he ended up deciding to go into another company. And as they were negotiating his salary, they asked him, how much are you currently making, which, of course, is the trap, because the next company wants to just kind of keep you around where you are and maybe give you a little bit more. And that is the trap question that keeps the income inequality for men versus women so prevalent. But he got that question.
And so he gave a very creatively high answer, which I don't necessarily recommend. But I think he said something like, I'm currently making $200,000. And he wasn't making $200,000. He said, I'm currently making $200,000. And so the response was, oh, there's no way in the world I can pay you $200,000. The best I could do is 175.
Is that OK? Which, to me, I would have never even dreamed of giving some sort of number that high, because I think in a lot of times, we don't value ourselves to be that. I'm not worth $175,000. You might be saying that to yourself. Well, maybe I'll give them a lower number, or I don't want them to think that I'm just too greedy, or something like that. So I think what helps in those types of situations would be to, first of all, understand the market, understand what a reasonable amount of money would be for that particular role for your area, like, a geographic area, and also for your specialty and for your years of experience.
And the worst that they could do is say no, right? I mean, you give them a reasonable number that is in line with a salary that you can justify, the worst they could say is, no, it's a little bit lower. But I definitely just encourage you to, like I said, just think a little bit more outside of the box and value yourselves, I guess, maybe higher than potential-- because they obviously have a vested interest in keeping your salary as low as possible. They just want you in the door in the cheapest way possible. So yeah, I highly encourage you to actually negotiate your salary. JEN HECHT: You raise a great point about the prior salary, because that's totally illegal.
I know it is in Massachusetts now to ask that question in an interview. I don't know how many other states have gone there. But that's the reason why is that it automatically kind of drags people down, especially those who are unwilling to get creative in their ask.
GABI DE QUEIROZ: One thing that I kind of learned to do is to have a network of people that I have some kind of like open space or open conversation, where I can ask the questions. I can say, I'm looking for a job. I have this amount of years of experience.
And how much do you think I should ask for? And they say, oh, I'm a senior person. I know that my company is paying around this. So I kind of like did a survey with my friends or the people that I know and asked them, what is the average or the interval that you think I should be asking for? The other good thing and tip is if you have any friend that is manager or hiring manager, ask them, because they know a lot about the salaries in the industry. JEN HECHT: Yeah. And the other thing I would say about that, too, is-- and you want to be careful with this advice, because if you're working in a very conservative industry, this might be harder to do.
But you should be able to talk with your colleagues in your current job about what you're making. And that can be very scary, but it's a good way to get information. And I think the workplace is changing a lot.
One of the positive things is, I think, more transparency. But if you have people that you work with in your same discipline that you trust, I'd encourage you to do that even where you are right now. And again, I've worked in a range of places. I think most places are kind of stopping this practice. It used to be like, oh, you cannot discuss salary, which I think also keeps a lot of people in the dark about what the real ranges are.
OK, I'm going to go to the next question. How do you decide which technical skills are worth the time investment to really learn versus having a passing acquaintance with? SYDEAKA WATSON: I think it would depend on your specific role and whether or not that is something that is needed for your job or for your growth. So I think about, for example, going back to that list of-- that massively long list of different types of machine learning tools and software and models and all these different kinds of things.
In your particular job, maybe you only need clustering. Like, they need somebody who's an expert at clustering and they need somebody who understands all the different ins and outs of a very small number of models. In that sense, maybe it's worth the effort of really digging in and understanding those. But the other side of that, obviously, is that what that does is that keeps you in the expert role for where you are now or maybe with the team that you're currently in. So another view of that will be to say, where do I want to go, or what is the type of job that I want to have or the type of career that I want to have for myself eventually? When I answer that question for myself, I want to be not just the best data scientist at my company, I want to be the best data scientist possible, so that any company can look at me and all the different types of experiences that I've had and say, wow, she really understands a lot of different kinds of tools, she understands that AI, for example, that's the buzzword of the day, and a lot of companies are interested in that, and so she's the person that I would want to call, because she has that type of expertise.
So I would definitely challenge you to think about your career and the path that you want to go on, and figure out whether or not some of these different types of tools will be useful. And some of that requires reading in the literature, but also looking at different use cases to see how other people are using it and the value that they're able to get from some of these different types of tools or models. JEN HECHT: This question actually came up in a session I was in earlier. What is the difference between a data analyst and data scientist? In your experience, is there a difference? GABI DE QUEIROZ: So in some companies, there is no distinction.
It's more like a title. But from my experience, data analyst is doing more querying the data, getting the data from a database, let's say using SQL, creating reports, where data scientists go one step further, where they also do the data cleaning, munging, and querying the data, but also is involved on the modeling itself. So that's my experience. But I think it's kind of like a gray area. Like, some companies, they call data analyst what I consider being data scientist. And they have different names, yeah, even more names than data analyst and data scientist.
JEN HECHT: Yeah. SYDEAKA WATSON: These are all squishy terms. So I guess maybe just thinking about some of the different roles I've had, people who've had the data scientist role, they tend to have a little bit more training, maybe a little bit more education, like, formal training in maybe like a math or statistics or physics or something like that. They tend to, again. Again, there's no hard, fast rule for any of these different types of things.
And they tend to have a little bit more expertise in some of the more advanced types of modeling scenarios, whereas maybe a data analyst would be very proficient in some of the data engineering tasks, or pulling different types of SQL queries, and maybe creating some dashboards or some reports. But the deeper insights of trying to understand, maybe starting with the question, of why are we losing so much money in this particular area of our company and this kind of product in our company? When I think of a data scientist, I think of a scientist, where you actually have to start from the beginning. Nobody's handing you a CSV or telling you specifically which types of data to look at or what types of models to use and so on. You're starting from scratch, and you're able to go through that entire scientific process and figure out where your hypotheses are, testing those, maybe validating those, maybe circling back and going back to the beginning. And so that entire process, I would think of that as that the person that has that type of holistic set of skills is a data scientist as well.
GABI DE QUEIROZ: And one last thing is data scientists, they usually make much more money-- SYDEAKA WATSON: Yes. GABI DE QUEIROZ: --than data analysts. SYDEAKA WATSON: There is that. GABI DE QUEIROZ: So if you are in doubt if you should write in your resume that you are data analyst or data scientist, go for data scientist. SYDEAKA WATSON: That's right. JEN HECHT: Yeah.
Well, it gets complicated, because in some companies, there's the official job title that's owned by HR, and then they'll let people make up business titles that say anything you want. So I've seen cases of somebody gets hired as a business intelligence analyst, and three weeks later, they're a data scientist. You're like, wow, that was fast.
OK. So a good question here about how do you handle being in a job with little or no mentorship opportunities? SYDEAKA WATSON: I kind of touched on this earlier. I've definitely been that-- how can I say it? I've been that person that has the set of skills, that one particular set of skills that nobody else in the company has.
So they literally cannot speak to me about it. And not to say that I'm inherently smarter than them necessarily. Everybody has their own specialty.
But maybe I'm the only one that has studied this particular area. And so I can't talk to them about those kinds of topics. And so what I have to do is create my own network. Just in the course of being a person who's interesting in developing my network in Dallas, for example, I go to different events. I go to different talks and just talk to people and find out what they're interested in, so then when I come across some kind of problem and I just like, oh, I don't know what to do or where should I go next with this, oh, wait a minute, David knows this.
He sounds like the kind of person that would have that kind of expertise. Maybe I can tap him. So that's one area when you think about technical mentorship. But the other path would be just growth for your career. So just figuring out, I want to be a director someday or I want to be a technical leader someday, how can I do that? And so some of that is just finding people who have gone through that path that you're trying to go through.
And that could be within your own company. That could be for people who work in similar types of roles as you're trying to get into, maybe who've been in different companies. So try to just be creative about finding people that can inspire you in the ways that you need. JEN HECHT: That's great.
David, I'm going to direct this one to you. Can you recommend any good Twitter accounts to follow to learn more about data? And already mentioned you highly leverage Twitter. What would you recommend? DAVID KEYES: Well, if you're not already following Mara Averick, that's probably the best place to start, definitely where I think I've learned more than anything.
Jeez. There's so many people. I'm trying to think of some other ones that have been particularly informative for me. I would say, this isn't an account, but the-- and particularly when you're developing in data visualization especially, the #TidyTuesday hashtag has been really helpful for me to see folks who are doing some really interesting data visualization. That's been a helpful one. Yeah, I think I'd probably start there.
JEN HECHT: Great. Anyone else have any thoughts? GABI DE QUEIROZ: What about the #rstats hashtag? DAVID KEYES: Yeah. Sorry. I assumed that was-- [LAUGHTER AND APPLAUSE] But yes, #rstats hashtag if you're not already following that as well.
The other thing, actually, looking over at you for a second, Gabi, makes me think that I really like the R-Ladies, @WeAreRLadies rotating curator, because it's really fun to see kind of each week. It's a different person who curates the account. And it's really fun just to see the different types of work that people are doing and the different packages they use. So I really like that account as well. JEN HECHT: Great.
So this question is more about R versus other things. So I'm fundamentally a generalist in my overall skills and my R skills, breadth not depth. Is that a viable way to build a career or is specializing essential? SYDEAKA WATSON: For sure.
JEN HECHT: It's valid, you're saying? Yeah. GABI DE QUEIROZ: No, like, this person is saying that all they know, it's R? JEN HECHT: They're saying they're a generalist in their overall skills and their R skills. And they're wondering, is that a viable way to build a career or should they be specializing within that, either in data science or in R, I guess? GABI DE QUEIROZ: Well, I think, again, there is a broad of, like, jobs.
So you can be a generalist or you can be a very specific. It all depends. I think there is-- maybe you have a better opinion than I. But I think there is space for everybody for being a generalist and to be a very, like, domain expert. SYDEAKA WATSON: Right. And I think I touched on this a little bit in the beginning when I was thinking about the different types of ways to classify the different types of skills that are required as part of a data scientist.
And so I was saying, if I wanted to just be a person that a general data science company would find attractive, then I would have proficiency in machine learning, programming, and big data technologies. And as our esteemed panelist said earlier, some of those companies might not necessarily care about some of those, right? So they might not care about the big data part. Maybe they only care about the other two. But when I think about it, again, unless you've already decided specifically, I know that this job in particular is what I want, and I know the specific set of skills that are required for that job, and I'm just going to keep on honing those skills, unless you're doing that, I would challenge you to branch out a little bit more, learn about other types of things.
But obviously, you don't want to get to the point where you're just picking up every single kind of tool or every type of model or expertise. At some point, you're going to have to specialize, because the people who have very specific domain experience, like baseball, like telecommunications, and so on, they really understand the data. So they can go beyond just saying, I understand that this column seems to be related to this column.
They really understand all the different ways that extra knowledge can inform how well you can get some insights from some of those different types of projects. So I would encourage you for your specific area, once you've actually determined that there's a specific area that is of interest to you, that you learn, you take the time to develop that domain experience. And the good thing about it is that if you get into it and you realize, eh, I kind of thought I would like this, but I don't like it so much, then you can get experience in something else. So yeah, just try different things. But I definitely would encourage you to have some type of specialization in some areas. JEN HECHT: Great.
We have a couple of questions sort of in this topic area about how you have dealt with being perhaps the only or first woman or person of color in your organization or team. Do you have any advice for people who might be, in addition to trying to get into data science, also dealing with issues in that area? SYDEAKA WATSON: So I have been in that scenario from the woman and also the person of color scenario in so many of the roles that I've been in-- actually, I want to say every role that I've been in. And yeah. And so one of the things that happens, commonly, is that as the one person, you feel like you have to be the representative.
If I mess up, then now all black people mess up, right, you know, that's just how we are, or all women mess up, right? So some things you just kind of have to forgive yourself and realize that you're human just like everybody else. You make mistakes just like everybody else. And just take those as growing opportunities to get better. But some of the challenges are external, some of them are internal.
So I definitely would just encourage you to realize, yeah, I'm a good person, too. I'm very smart. And I have a lot of great ideas and good things to offer. And so just kind of keep that in mind as you're growing your career. GABI DE QUEIROZ: Yeah, also, don't feel like you have to be the one responsible-- like, let's say there is no diversity efforts and then you are going to be the one doing the whole work, because-- so, like, try to get other people to help you as well, to have allies. And also, the outside network is very important in those times, so you don't feel lonely, so we have this support around you that keeps you going.
JEN HECHT: Great. So all of you had such different paths. But we have a question here about, for a young data scientist trying to get their first job, should they be looking at large companies or startups? Do you have any specific advice about what context might be the best place to start? GABI DE QUEIROZ: So I always worked for startups until moving to IBM almost two years ago.
So I went from a company where we were 20 people to, like, 300,000. It's a big jump. I don't know what is better. I think if you are a beginner, the best company is the company where, again, you are going to have support mentorship, because it's so hard to break into this career without having someone mentoring you.
So if there is a startup, that you have a good network and support, yeah, go for it. But if you have opportunity to work for a big company, you probably have more mentorship. But also, like, you are going to be more specialized in a startup, where you have to do everything from getting the data, cleaning the data, data engineering, modeling, and putting protection, for example. DAVID KEYES: Can I also just give a pitch for the idea of potentially considering consulting, because I think it's a really good way that you can gain experience.
And even if your ultimate goal is to become, say, a data scientist, I think there are a lot of opportunities to use the types of skills that you have as an R user outside of that title in ways that can really be impactful for the organizations that you work for. I mean, just as a minor example, one of the clients that I was training recently was working, doing an evaluation of an after-school program. And they were working at 100 different schools and they had to produce literally 100 reports. I taught them about parameterized reporting and it just completely blew their mind, the idea that they could automate that and not have to manually produce 100 reports.
They're not a data science firm. That's not how they identify themselves. But those types of skills can be really useful and serve a wide variety of organizations. And you can do that sometimes through consulting. JEN HECHT: I actually have a follow-up question directed to you, David, on that point, which is, how did you get started consulting? How did you start out finding clients? And how has that changed over time? DAVID KEYES: So I started out consulting-- the honest answer is because I had twins three and 1/2 years ago and I wanted more flexibility than in the job that I had at the time.
I started out finding clients-- I've mostly done online marketing. I offer online courses as well as trainings for organizations. And I've tried to basically become a resource for other folks. So I write a lot of blog posts really pitched at newcomers learning about packages to make effective tables or that type of thing.
And that's really been my most effective strategy to reach out to clients. JEN HECHT: That's great. So I think we're right at time. So I'd love to give a round of applause to the panelists. [APPLAUSE] Thank you. [CHIME RINGING]