Speaker 2: (00:01)
Yeah. Humans just be out to go out there, ask questions. If you want to do something, ask your boss and do it. If you don't have that place and that table, ask for that. Be proactive cause I feel like it's just a limited million that we put to our minds. And then this is how people see us. So if I say, you know, I'm going to be a leader and let's say it's, I'm going to work hard for that. Nobody will stop me. 

Speaker 3: (00:22)
(Background music)

Speaker 3: (00:30)
(Background music)

Speaker 1: (00:44)
what's up everyone? Thank you so much for tuning into the artists of data science podcast. My goal with this podcast is to share the stories and journeys of the thought leaders in data science, the artists who are creating value for our field through the content they're creating, the work they're doing, and the positive impact they're having within their organizations, industries, society, and the art of data science as a whole. I can't even begin to express how excited I am that you're joining me today. My name is  Harpreet Sahota   and I'll be your host as we talk to some of the most amazing people in data science. Today's episode is brought to you by data science dream job. If you're wondering what it takes to break into the field of data science, check out dsdj.co/artists for an invitation to a free webinar where we'll give you tips on how to land your first job in data science. 

Speaker 1: (01:36)
I've also got a free open mastermind Slack community called the artists of data science loft that I encourage everyone listening to join. I'll make myself available to you for questions on all things data science and keep you posted on the filing fee open office hours that I'll be posting for our community. Check that out @artofdatascienceloft.slack.com community is super important and I'm hoping you guys will join the community or we can keep each other motivated, keeping each other in the loop on what's going on with our own journeys so that we can learn, grow and get better together. Let's ride this beat out into another awesome episode and don't forget to subscribe, follow, like, love ,rate and review the show. 

Speaker 3: (02:19)
[background music]. 

Speaker 1: (02:33)
Our guest today is a high energy data science professional with a proven ability to rapidly prototype new machine learning techniques and delivery 

Speaker 1: (02:43)
on feature requirements. She's an innovative data scientist who's coupled her visionary and conceptual thinking approach to develop statistical and machine learning models to create scalable NLP solutions that can be embedded within larger systems. She's got a knack for crafting clear arguments, stunningly coherent presentations and uses for powers of persuasion to listen, influence, and network effectively throughout her career. She's gained significant work experience in industry and applied research and as attained a series of academic accolades including bachelor's and master's degrees in telecommunications engineering from the Polytechnic university of Toronto, as well as a PhD in computer science from the university of New York where she developed a new algorithm called randomized greedy ensemble outlier detection with grasps say that three times fast. She's worked at organizations such as the Canadian Institute of technology and click. She's currently at diva systems where she develops models that leverage NLP and machine learning to help drive meaningful insights and quickly identify, classify and prioritize adverse drug events. 

Speaker 1: (03:45)
She's also a data science mentor with sharpest minds where she mentors and guides, mentees in STEM fields that want to learn data science and NLP. She's a well known and respected member of Toronto, women of data science where she lives out, her passion for giving back to the community by helping and advising women break into the field of data science. So please help me in welcoming our guest today. A woman whose recent honors include five published papers across various academic journals. Dr *** led the owner in a shiny nice. Thanks so much for being here today. Let me own it up. Uh, really appreciate your time. 

Speaker 2: (04:15)
Thank you so much. Harpreet it and thank you for inviting me for giving the opportunity to, uh, talk about my jury knee and the splash for and just, I dunno, inspire people and what it, what a good thing to do here today on a saturday, just talking about and just having fun together. 

Speaker 1: (04:34)
So, so happy you were able to carve out time from your schedule to come here and talk about your journey. You come from an amazingly strong research background. So can you talk to me about your journey from the research world to data science and touch on some of the challenges you faced along the way and how you overcame them? 

Speaker 2: (04:52)
Yeah. So, um, my, um, my career journey is not linear at all. So I started, uh, like finishing my master degree and then I was appointed as a lecturer in university and then that I kind of jumped on, uh, upon getting, uh, pursued my PhD degree in computer science. So from telecommunication you go into computer science with no computer science background whatsoever. Then like, I just read that opportunity, uh, and then, uh, learn to learn to law. So, um, you know, I was lecturing as well and building my CSEs and researching or publishing a lot of papers. Uh, it has been a challenging but rewarding as well cause that was my first exposure to a data science. So from my, uh, from my, uh, advisor, my thesis advisor and, uh, so I learned, you know, programming, I learned machine learning at the same time, reading a lot of papers, getting to understand algorithm under the hood as well. 

Speaker 2: (05:56)
And then, yeah, papers it was kind of challenging the same time. You need to revise them and then, but to learn in the beginning perhaps you don't know how to write a paper and then through virus feedback you just apply them and then just you, you end up publishing in high impact journals, which was really rewarding. And then after that I um, you know, I did a separate two. We needed a, I sound like I just needed more hands on experience to apply those kind of principles and all this wealth of knowledge into a business problem. So I just decided to migrate in Canada, here in Toronto. So, uh, and then here is, where was another shift, another pivotal moment where I kind of applied myself in the industry. 

Speaker 1: (06:44)
Iteration is such a, such a pinnacle part of data science. I almost feel like it should be called iteration science. 

Speaker 2: (06:52)
That's true. Experimenting, experimenting and figuring it out along the way. 

Speaker 1: (06:56)
So your area of focus, that the thing that you're really passionate about is NLP, which is such a fascinating part of data science. To me, it's kind of like black magic. Um, can you kind of talk about what got you interested in NLP and what you think the future holds for this particular area of data science? 

Speaker 2: (07:15)
Oh, well, I just, you know, I just ended up by chance. Uh, honestly, cause I, I came here in Toronto, I was looking for jobs opportunities and my, my thesis was in Larry detection and just a, a building, uh, greedy ensemble algorithm for applied detection so you can begin seeing codes that previous experience into time series problems. But just, I just ended up in a, you know, early stage startup. They were working for a building their own checkbook. So that was my first exposure to animal P. so I started learning, understanding their data, text analytics, running the topic modeling algorithm. So I just started learning, really kind of beat up these stats. And then after that I was a, I went to another company where from my experience in the topic modeling and NLB, I was kind of associated, so that NLP salt say in the team where I had a really nice or really helpful, uh, coworkers that I learned a lot from them. 

Speaker 2: (08:14)
So I joined their a research problem and that is where I started. Uh, you know, my passion for NLP started to, uh, to expand. It's not that I had something you need, but I feel like passion always, uh, get to develop. When you start working on that, you iterate, you fail, you, you, you succeed after that and then you start liking it. So yeah, in terms of a mop, as a field, it's growing so much. I'm learning every day. I cannot say that everything. So it's an ongoing process and I, but what I love about it is to make sense of unstructured text. So I'm working now with first events or reports or like clinical data, medical data, which is so dirty. I, so and so on structured, just being able to prize that, to understand the connotation, the meetings, the relationship between different medical concepts. 

Speaker 2: (09:13)
It's so fascinating to me. And in terms of the future, like the future is that there'll be a do this is uh, this is my, uh, my 2 cents. So, uh, it will be a huge thing. And to business for understanding of they said, Oh, voice to text, speech recognition, uh, love. There is a lot of room for that. And what I love about it is the side that how can we understand human language, how can we make computers understand human language through different nuances from food, different mannerism or like we use different words with a different context, these on the time or the concepts of studies used. So being able to meet computers understand that, I believe this is really fascinating. So we are, you know, on red track for that. 

Speaker 1: (10:03)
Funny, if my computer started to understand and you and I, when I'm cussing at it because my code doesn't work, so let's hope it doesn't understand me enough to, uh, to use it against me. You know, it's fascinating. There's a comment you made that I, that really resonated with me. Um, and that's about passion and how passion is not, um, you develop your passion, right? You develop working on hard, tough problems. I think that's a very important point that you brought up. So, um, reminds me of this, this book that I'd recently read, uh, perhaps he might've read it as well. Uh, SO GOOD THAT THEY CAN'T IGNORE YOU by Cal Newport. 

Speaker 2: (10:35)
Oh, I haven't, I'm, I'm putting that in my bucket list right away. 

Speaker 1: (10:40)
It's a great read. So now as somebody who's kind of been on, on both sides of the fence, 

Speaker 2: (10:44)
yeah. 

Speaker 1: (10:45)
Talk to me about some of the common challenges you've seen up and coming data scientists face when it comes time to take research into production. 

Speaker 2: (10:55)
Yeah, it's a, it's, it's challenging I would say. I would say cause uh, research, uh, you have the kind of time or air cover I would say that you have the time to deep dive into that to make it perfect, you know, find the, the, the latest with surge, combined them together, be creative and you can find the best and the more we're trying to solution. But on the other hand, in the business perspective who don't have that time, you need to have a solution that it would be visible that that has and that it works and then you can build upon it. So in terms of the members, I knew when I was working, I just did a lot of research. I like six months, a really great project. However, uh, we went to just making it as a proof of concept and we didn't have the data to make it, uh, to make it like retrain our model and Nico productionalized code and build the app because we didn't have the clinical norms

Speaker 2: (11:53)
So sometimes when you work on research, you can do really great job, but uh, what is the value of it? That is something that you need to be, be mindful of when kind of throwing up, throwing yourself out there. Especially for, for for business problems in terms of like, uh, there are a lot of labs that are, you know, pushing the seed off to the research, building new algorithms. That is another story. But in terms of, uh, it's kind of a nice match and I'm facing that kind of challenge cause I'm used from research groups company and now here a, they require time, the speed that they require, the deadline to release the screens. So you need to create that kind of meaningful viable product and ship it and then perhaps you might build upon it and make it more sassy or that is a like a challenge that I see. And then on the other hand, coming from the research background, we are used to, you know, feel that algorithm but not really looking at the code or being able to write to, uh, to productize the code so we can, uh, put it into a, into production. So I'm learning that as well now, which is a great experience. Collaborating with the engineering team and then getting their kind of coding standards that I wasn't able. I wasn't used to before. So it's great. Yeah. 

Speaker 1: (13:16)
WhatŐs up artists? Check out our free open mastermind Slack channel, the artists of data science loft at art of data science, loft.slack.com I'll keep you posted on the biweekly open office hours that I'll be hosting and it's a great environment and community for all of us to talk all things data science. 
***Background sound

Speaker 1: (13:41)
yeah. Somebody who's curious, who has a curious like myself. Like, it's, it's such a curse sometimes when you want to go down that rabbit hole, but then you at your boss like yo, like we need to, we need to get some out there and he'd start making moves and you know, adding business value 

Speaker 2: (13:58)
Because you want to narrow down, you want to like a research a, at least for me sometimes I, I find myself, Oh I want to read as much as I can. To me,  knowledge or what have other people donk know out there. I want to to get the latest research lead us algorithm and combined and then okay, but like we don't have time, we need to start a diplomatic, then we need to start doing that. 

Speaker 1: (14:20)
So, so what's the first thing you do when you're taking on a new project? Like what, what are some of the steps you take to kind of keep you on track while you're navigating the ambiguity of some of these, these projects? 

Speaker 2: (14:31)
Yeah, great question. So always I start with a problem. So I always start with understanding of the problem. Uh, so what, what is the time? What is their time asking, what is their want, if this is an internal project, was it what is required from me? So, uh, and trying really make it crystal clear in my head first. And if I have any doubt, I always ask my colleagues, never assuming I find sometimes I assume I'm wrong. So it's always to have this clear vision of what will be the problem and then think about what is the result that they're looking for. And then kind of reverse engineer out with that. So let's say, uh, identify the KPI key performance indicator. What are the results? What will make this project successful? And then after that, think about, you know, okay, let's go translate this business problem to machine learning solution. 

Speaker 2: (15:30)
Here you can have some research, but always with a specific timeframe and brainstorm and think about a solution. Always start with a baseline. Always start with the easiest algorithm that works and then you can build upon it never started. Uh, I have done my other state, the like always I wanted to do something complicated, different, difficult or stuff like that. And just it didn't work. So just do something baseline and then you can improve it and always look at the data and who don't have the data. Always convince your boss having this kind of procedure. And I believe it's really crucial. Get the data, look at the quality of that, label the data as well, if it's needed, and then realize it and deploy it and apply it to productions.

Speaker 1: (16:21)
Yeah. That's such an important point there to create a baseline, uh, because you know, you create that baseline and that kind of comes the new benchmark, the score to beat. Right? So you know that if, if whatever you do doesn't perform better than that baseline and you're headed down the wrong path, yeah. Go down the rabbit hole. Right. So it's important to have that baseline as, as a North star. Besides the obvious technical skills that are required to be successful as a data scientist, what would you consider to be an essential skill to be and remain successful as a data scientist? 

Speaker 2: (16:54)
Uh, I believe in this time we need communication. Like a communication is the key to be successful cause you need to communicate with different stakeholders. Like in my position right now, I need to communicate and collaborate with PMs, product managers. I need to understand them. I need to understand their language. Uh, on the other hand, I need to communicate and collaborating with the engineering team, with software developers, which they have completely different angle, all the problems. And you need to be able to convince this oldest equal stakeholders for your solution. So you need to be able to talk in their language. So this is really crucial because they are not convinced that the solution is the right one. You don't have other support, keep, keep the project going. Uh, another thing is apathy. I do, I, I just, uh, I'm really advocate of empathy. Like looking at the client, what they're looking for. So what is there one? So putting yourself in their position, putting ourselves in the social engineering team, what is their will they are looking for always miss you kind of need your communication easier. And if you have, let's see, constantly. So or like always thinking and they're being in their shoes, really, it will help you to kind of smooth it out and just, you know, again that kind of approval and yeah, make yourself more successful in the team. 

Speaker 1: (18:25)
Awesome. Yeah, no, that's really the emotional intelligence is definitely something that I feel like a lot of a lot of people don't put, you know, work into. Like a lot of the times we're focused so much on learning, you know, the new and new algorithm we're learning some new programming technique. Um, but you know, putting in the work to really understand where your fellow humans is is an essential skill as well. So I 100% agree with that. Um, so I know we kind of, you know, we kind of touched on on this, but piggybacking off that last question, uh, what are you looking for, uh, up and coming data scientists? 

Speaker 2: (19:02)
So yeah, uh, when I, uh, we were hiring some interns at a, at our team and I was looking for is first being curious, like being critical, thinkers  And what I look for is like let's say you give a problem to them and they haven't seen it before. They should be comfortable or like figuring it out, asking questions and giving, giving their assaults and even though they might know it, they might don't know it because they haven't done it before. They should have the kind of internal confidence. I'm going to figure it out no matter what cause asking questions, bouncing of ideas, a white-boarding, these are skills that you don't find it and this comes with experience as well. This kind of confidence- and what else? So I see like being lifelong learner as well and I feel like being resilient, this is like one of the key, key things that I see. It's hard to test in the interview but I feel like you build it through experience that's in your cell phone. We have so many projects. Keep going, keep going and learning from them and then apply to the next projects. 

Speaker 1: (20:15)
That's that's a great segue actually into my next question here. So, I mean, apart from your, like just your amazing technical skills, what do you think is, is the one quality that's really contributed most to your success? 

Speaker 2: (20:29)
Yeah, this, yeah, it's related to honestly, like I didn't start with really strong technical skills. Uh, as I said when I started my PhD, didn't have a coding skills background at all honestly. And I learned along the way and I seem like having this growth, it helps me a lot. Like, you know, uh, even though I didn't know it, I'm gonna learn it no matter what, not comparing myself to others cause I'm having my own journey and just comparing myself to yesterday learning am I doing percent incremental, um, incremental growth, incremental improvements to myself. And then I'm good and I'm always, I have this, uh, being comfortable, not being uncomfortable. So in the sense that I, for myself, whatever it is, I'm going to grow from this. Nothing, nothing bad will happen. Even if they reject me. Let's see, in an interview or something, I will get something out of it no matter what. So just just do it never stops something cause people gets, you know, procrastinate. They have the fear of rejection. I'm always going to write the book. I'm not gonna, I don't know, feels that a model cause I know equal sale, they kind of had this limiting belief stopping themselves from this. So I learned that. Just do it. Just go out there, throw yourself into the sea and just learn. 

Speaker 1: (21:52)
I literally, I have chills all over my body. That was so powerful. Right? Having that growth mindset is so crucial and I think it's that one, like you mentioned that one belief that that matters is that all of your efforts, everything you put in all that hard work, even if you don't get the outcome that you want, you're still going to be rewarded with the new knowledge, new skills, new abilities. So it's always worth putting in the effort even if failure is a real possible outcome. 

Speaker 2: (22:18)
Yeah. Yeah. And to add up or like I was talking with my friend the other day and two months ago, she was looking for a new opportunities for a better position and she was having some rejection in terms of interview. And I told her, keep going, you know, keep going and learn from those rejections and learn from that, from those interviews. Are you better than the next teacher? Yes, you're good. And then she got her job, her dream job because the thing is the universe will conspire with you after you do your hard work, your dreams of will come along and then it will be easy then because you are ready for that. 

Speaker 1: (22:53)
True words have never been spoken. So I was wondering, I was wondering if you could, you could speak to your experience being a woman in tech, your involvement in Toronto WIDS and if you have any advice or words of encouragement for our female listeners. 

Speaker 2: (23:07)
Yeah, so I'm pretty passionate about like, you know, empowering women, uh, you know, passing that to, to the community as well, helping them. So, um, my experience with women in data science that's around is pretty interesting cause like I just, I just gave me in 2017 and I just got this message from LinkedIn or the organizer of the conference, perhaps I was the PhD and the Q word. Got it. And then she just invited me to the conference. I didn't have any experience whatsoever talking to a Canadian audience and they said, okay, I'm going to go, you know, I just like to entertain the audience. I don't care about my accent or anything like that. I'm just, and I just presented thesis and it was a huge success. I liked it. And then the next year again, she, uh, she asked me again and I, uh, uh, presented my research work that was grown after one year. 

Speaker 2: (23:59)
So, uh, and in terms of, yeah, an advice that I would have to women, I just feel like don't have the limiting belief that you are a woman and then you, uh, I don't know. You will kind of stigma and all that just, you know, you are the same manner, the same. They have their own insecurity and they fell down themselves. We are humans. Just be out and go out there, ask questions. If you want to do something, ask your boss and do it. If you don't have the place in the table, ask for that. Be proactive. Cause I feel like it's just a limiting belief that we put to our minds. And then this is how people see us. So if I say, you know, I'm going to be a leader and let's say it's, I'm going to work hard for that. Nobody will stop me. 

Speaker 1: (24:48)
Love that perspective. So before we jump into our lightning round here, what's the one thing you want people to learn from your story? 

Speaker 2: (24:55)
So I just want to inspire people from all my journey and just don't be afraid. Just be fearless and try things out. Just nothing will, even if you fail, it's okay. It's a learning that's their cell phone. And go again to your goal. Just stay focus. And at the end of the day, I just came from Albania, from academia and all this kind of differences. And I don't know where I will end, honestly. It's a long journey. You learn and you learn every day and then just be faultless and just be resilient. How do you learn resilience? How do you build that muscle by failing. 

Speaker 1: (25:44)
Are you an aspiring data scientist struggling to break into the field or then checkout dsdj.co/artists to reserve your spot for a free informational webinar on how you can break into the field that's going to be filled with amazing tips that are specifically designed to help you land your first job. Check it out. Dsdj.co/artists

Speaker 3: (26:06)
***Background sound

Speaker 1: (26:09)
And you know, in the short amount of time that I've known you, I could tell with that attitude and just the tremendous amount of work that you've always been putting in that 

Speaker 2: (26:17)
yeah, 

Speaker 1: (26:18)
it's going to be at the top or beyond the top. 

Speaker 2: (26:21)
Yeah. Well thank you. Hope so 

Speaker 1: (26:25)
let's jump into our lightning round here. So Python or R? 

Speaker 2: (26:28)
Uh, honestly I know only uh Python and uh, I can, is the good to go . ItŐs, you know, the most used 80% to learn and just be better at it. Code every day code every day. So even if you have those bugs, they do need to debug it and you are saying, Oh my God, I'm not finding it. People ask the, your your support group and yet we will see you right now. 

Speaker 1: (26:51)
What's your favorite algorithm? 

Speaker 2: (26:53)
That is pretty question to me. I would say like my algorithm that I (chuckles)É is my algorithm.. , I just feel like it depends on problem that I'm looking for. Now with medical data, like just using support vector machine for a classifier, it's works wonders and generally with NLP now I use SpaCy, uh, as a library to get the dependency of features. So it depends on, it depends on the problem. Like I can't, I don't have one one favorite one. 

Speaker 1: (27:35)
Uh,  it's, it's the one you wrote. It's okay. 

Speaker 2: (27:38)
Yeah. 

Speaker 1: (27:41)
What's a book that every data scientist should read? 

Speaker 2: (27:44)
Uh, I have two favorite one, one I love and one, uh, more, uh, more, uh, algorithm. So the first one is the Data Science For Business. Uh, that is like the book. It's a jen, honestly, for everybody that wants to understand how are we solving business problems and just in a high level perspective, it's a really great read for managers as well is they want to manage that data science. It's a great read. And then for uh, for in terms of understanding algorithms under the hood, uh, their principals, their maths. I, I like the hands on machine learning with psychic learn and test on flow I find is really helpful for, uh, crucial, uh, questions or an interview, let's say. How does the support vector machine, uh, works under the hood or, or around them forests or a assembles models like you have a really interesting and really helpful insights of how they work and yeah, these are my two in terms of like for women, I love the essence of book, forget the man and get a sponsor. That is a really eye opening book for me and this can help you kind of like climb that career ladder faster. 

Speaker 1: (29:00)
Sylvia Ann Hewlitt, right? 

Speaker 2: (29:03)
Yeah, I guess so. Yeah. 

Speaker 1: (29:05)
Yeah, definitely. I'll, I'll be sure to link those books in the show notes. How about a book recommendation for people that are wanting to learn NLP. 

Speaker 2: (29:12)
You can start from second learn feature extraction. That is the fire that I started with TF IDs and the vectorization. 

Speaker 1: (29:19)
So what's your favorite question to ask in a job interview? 

Speaker 2: (29:23)
I love the question. My favorite question is asking to the manager. So I can gauge what kind of a culture they have. Are they motivated, are they passionate about what they do? They have a sense of purpose. So I always ask, so what makes you wake up in the morning and go and go to work and just keep all that day? And that is where they are so passionate and just, they kind of explained their purpose. I can kind of gauge these. They're not a reply with no, that answers the ASM. I feel like perhaps this is not the right job for me. 

Speaker 1: (30:00)
Yeah, that's a really, really good question. So what's the strangest question that you've been asked in an interview? 

Speaker 2: (30:06)
Uh, what, what is your spirit animal? It's not a strange question. It's a, it's a good question. When they asked me, I felt like, okay, you know, so I would say like I may be a Chad but I might be a tiger. Yeah. So that was interesting cause that is kind of aging or culture fit or not. So how do you see yourself? So that's time a cat. But when I see myself in the mirror, I am tiger or whatever. 

Speaker 1: (30:36)
Nice. It's true man. My spirit animal is actually a Fox will because there's a, there's no old adage that difference between a Fox and the hedgehog. The hedgehog knows one thing really well, but the Fox knows a little bit about everything. I just feel, yeah, generalist. Something like that just resonates with me. How can these people, how can these people connect with you? 

Speaker 2: (31:02)
LinkedIn, I have a big network in LinkedIn and I always reply to people while they want to pick my brain or even if they want to talk to you for mentoring. Uh, my LinkedIn is ***. You can find me on LinkedIn. 

Speaker 1: (31:18)
Yeah, I'll be, I'll be sure to link your profile link in the show notes as well. Oh yeah. Seniors thing. We want to have on my phone. I got rid of like Facebook and Instagram from my phone and just kept LinkedIn just because it's got all the information that I want to be exposed to. Lydia on a, thank you so much for being so generous with your time today and I really appreciate your insights and your knowledge. I know a lot of our listeners are going to benefit from everything that you had to say today. So again, thank you very much for being on the show 

Speaker 2: (31:47)
And thank you so much for inviting me for giving the opportunities to share my stories and this, uh, this platform. So I wish you all of us in your, uh, journey. 

Speaker 1: (31:57)
Thank you. 

Speaker 2: 
Thank you so much.