Shrish Mishra did his M.Sc. in physics and then he was doing his Phd in astronomy and he gained interest in data science. Then he starts learning data science skills through online resources and then he got two job offers as a data scientist. Right now he's working as a research data scientist at The Math Company. In this video, we are going to go through his entire journey. Towards the end we will discuss two important things: number one if you are from non-computer science background and if you are not getting interview calls for data science career, we will discuss some effective networking tips. And the second thing we are going to discuss is what kind of tips you can use to make this transition more smoother. Let's get started! Shrish Mishra thanks for spending time for today's conversation. Let's start with your current role. What do you do and what kind of technology and tools you use in your current role? Yeah first of all thank you Dhaval for having me here. I am very happy to be here, uh you have been always an inspiration for me. So yeah, so currently I am an associate data scientist at The Math Company, so this is a analytics consultancy company. So in The Math Company I can call myself a kind of research data scientist. We implement some state-of-the-art uh uh algorithms for our like in-house product. So my basically my job is to read many papers and implement some of them in mostly mostly python. So yeah we work mostly in python for now. Research data scientist sounds pretty cool. I have interviewed a lot of folks who are data scientists they use ready-made tools python, BI tools and they just solve commercial problems. But your role is more into research so it's kind of cool. You like the the the research work? Yeah yeah this is this is the thing that I liked most here because I was doing research in physics before coming here. So this is the thing exactly that I was doing also. So this profile actually fitted me well I can say yeah. I see. Okay. So then let's talk about your background because you have a degree in physics. So let us discuss your education and then we'll discuss what got you interested in data science. Uh yes of course of course. So actually I got my bachelor's in physics from University of Delhi and after that I went to pursue integrated PhD from Niser Bhubaneswar. Uh Niser Bhubaneswar is an institute by department of atomic energy. This is a research institute uh and I was into integrated phd uh before I came to know about data science, and then I when I decided to switch into data science, so I had to leave my Phd and on leaving they awarded me a with a MS degree. So now I have a master in physics degree. And yeah during in my Phd uh I was working in astronomy, my topic of work was astronomy. Nice, astronomy sounds so cool actually because I go to local planetarium here and I like doing stargazing and things like that. It always fascinates me, so I'm I'm really happy to talk to someone who is a formal education in astronomy. So here you are you're studying your physics, astronomy and then all of a sudden like something um sparks an interest in data science. So we want to discuss that like how you got into this data science field? Of course, Of course I would like to I'd love to discuss about that. So once in a conference I went and there was some person who has worked on some machine learning algorithm on astronomy datasets. So in that talk itself I grew my interest in data science and further luckily I went to the another conference and there I had a chance to work in a project in which I had to work with machine learning algorithm in astronomical datasets. So my job was basically to classify star images with their spectral data. So this for for doing this I had to read some something about machine learning algorithm. Ah before that, I also used to think that machine learning is mostly computer science topic. So I am physics guy and I will not be able to do that. But uh very very soon I understood that this is just modeling and statistics and this is what we at the physics have been doing like since very long. And yeah glad you mentioned that because many people have this misconception that they need computer science degree to go into data science field. But they don't realize that data science is just a tool which can solve problems in your domain. So if you are a mechanical engineer or electrical engineer or physics person in your domain, you can use data science as a tool to solve variety of issues like like you know how you solve whatever star classification problem. So when you started working on these projects did you know python at that time and if not how did you learn python and any other technology that you used to solve that that problem machine learning problem? Yes uh yeah I knew python quite a bit because I was working with python in my project itself. So in astronomy like lots of people work with python. So I also like had a good experience with python. So picking up python was not a big problem for me. Yes, so python I knew then I read some of the books, some of the blogs, some of the video to understand over all of the machine learning and data science. And yeah at that time applied I applied that algorithms there. But later when I came back I started reading and growing more interest on this. Got it, this is interesting so you going to conferences is very important. I think people can learn from your experience that when you go to conferences you meet people who are solving issues, you know different type of issues and projects and then you all of a sudden realize oh I can use data science in astronomy! Is that right? Yes, that's absolutely right, and you also mentioned many times in your videos that you should start applying machine learning and data science wherever you are right now. Exactly yes that's that's exactly what I say that whatever your field is, whenever there is, wherever there is data there is data science. So in your current field if you're dealing with some data let's say you are HR you are looking at like employees salary trends, things like that start applying machine learning because that is a based and a natural way to transition into this data science domain. Great! So you know python, you of course I think you knew statistics and mathematics right quite a bit as a physics student? Yes that is right, I was trained quite a bit in statistics and mathematics. Linear algebra I had like it was very well mixed with me because in quantum mechanics you require a lot of linear algebra. Very good. Yeah, so linear algebra was good. yeah, so then now okay so you have good mathematical statistics background, you know physics, and you find a problem in physics domain which can be solved using data science, so how did you learn machine learning and any other things? I know you knew python, statistics, math but then machine learning and any other things like jupyter notebook or any cloud solution, how did you learn it? Did you get any degree or did you learn everything online on your own? Uh yes so most of my learning came online. So I took help of lots of free resources that is already out there mostly from youtube and some from Udemy one or two courses I would like to mention that is from Udemy. So I I you I read a lot of book, I like I like reading books instead of watching lecture. I don't know that is how I am. So uh I can I started with a book called hands-on machine learning. So that is a very famous book by Aurelien Geron I think. So uh this book has like nice combination of code practices and some like quite good theoretical description is also there. So this is very like this is where I started. Uh later uh I did a book uh which name is python for data analysis. This is by Wes Wes Wes McKinney yeah. So he is the same person who wrote Pandas library. Yes, yeah so that book was nice that was enough for like me to qualify as a data analyst analyst and later for a little deeper understanding, I read a book called intro must intro to a statistical learning by Robert Tibshirani that is a very nice book. He has given it free for everyone and and he had also had some lectures in youtube, so yeah that was very helpful and of course guidance from this channel codebasics was very helpful for me. Like whenever I was stuck somewhere I came here in this channel and I got something out of here. So any guidance, anything, any small thing, any big thing yeah everything was there in this channel. I am following this from like beginning from my journey. Oh I see, yeah thank you. Well thanks thanks for your kind words that is really nice of you. We'll provide links of all those books that you mentioned in the video description below. So in your journey here you are where you got this project in your domain, now you learn machine learning, statistics all of those things using books and online resources: youtube videos and those udemy courses and so on. I'd like to name one person uh I think his name is Mike X Cohen. So his course on statistics and linear algebra is awesome. So he has given course with a lot of python code. So you can practice there and also he has explained the theories very well. So yeah and one more thing I would like to add like this is this is very important. Uh the thing the most closest thing in the college that we we study that is even closest to the data science that we do in the industry is econometrics, you must have been knowing that. Economic econometrics, yeah econometrics so there is a lecture on econometrics by Ben Lambert. So his lecture is nice, he has explained everything linear regression, p value some statistics, and also time series analysis. So yeah uh that course like you should be you should consider taking anyone who is like uh trying to come into the data science. Great, we'll provide the links of all the courses in the video description below. So people can refer to it. Great, now now once you learned all the skills and when you worked on your project in in physics domain, now what was the next step? I think by that time you made up your mind right that you want to become a data scientist now. Um yes. So then yes I I mostly made up my mind. So but yeah uh coming out of from Phd and getting a job in data science was like quite difficult for me. In my institute uh environment is mostly like research in research environment, most of the people go to U.S. for a Phd degree. But probably I'll be the first person to go into data science. So yeah it was very difficult but luckily I got I got the support of a mentor, uh his name is Vishwajit he is my friend like in LinkedIn Yeah he like he like told me many strategy little little things how to search for job another. So how did you find this mentor because I have mentioned this in one of my videos that finding a mentor is very important. So is yes so is this person Vishwajeet is he your senior or you just know him through LinkedIn? Yeah I just know him through LinkedIn? I actually reached out a lot, I tried to ask for help for everyone who is there in LinkedIn I faced many rejections, many people ignored. But I keep continuing and actually yeah you just need one help. Okay, so you looked how did you find research so you looked for just data scientists, okay data scientist in let's say India something like that or how do you do? Yes, yes yes I usually used to search data scientists in India and I used to send them a request, connection request and after they accept the connection, I send some Hi, and something like I'm interested in going into data science I'm physics I'm from physics this and that and yeah some people reply some people doesn't reply. Yeah but most of the people reply. Great, so now you found this amazing mentor who provides you all the guidance so then how did you start searching for job and what was your resume preparation process? So let's cover your resume. So in your resume what kind of projects you mentioned? I think you already worked on this machine learning project so maybe that was the project and did you mention any other projects in your resume? Uh yes that machine learning project was there that I did my during my Phd and one of the one of the one more project was there, so that is that was also like I pursued that during my Phd that was image classification project using some other mathematical techniques then deep learning and some of the project I did externally from from Kaggle and from UCI ML dataset. So uh similarly like I presented four or five best of my projects there in CV and I mentioned what my studies are what my uh like what are the subjects that I'm good with these kind of thing I mentioned there and yeah. Nice you so you have now nice project portfolio. Of course your mathematics background helps in in data scientist job. Uh now many people ask me this question that if you don't have computer science degree you don't get an interview call. Is this the problem that you also faced and how did you came out of that problem? Yes that is kind of true. So like yeah most of the people want computer science scientists like their uh data scientists, uh but there are also some people in their market uh who wants maths and statistics people as their data scientist. So I also I also had like chance to uh got interviewed uh by a person. So he told me personally that he is looking for someone who is good with math and a statistic. I think if you have math and statistics you have already an edge like you have some little edge but you have to reach out a lot Yeah that is what I would say. Got it, so then how did you reach out to people? How did you, did you just go to LinkedIn job and start applying the jobs with the resume that you prepared, or you use some other strategy? Yes so initially I just went to the LinkedIn and I applied blindly. I used to wake up in the morning and I used to apply for 10 jobs, uh like at least 10 jobs whatever there I used to blindly apply that. But let me tell you it didn't work at all, nobody called me. So I then changed my strategy, so in my like uh another strategy what I did that I I gathered some of the email id from a startup india, crunchbase and from product hunt, from zauba corp so I collected some of the emails, I started sending some personal emails to them with a nice cover letter. This part is important you can't just attach your CV. Nobody has time to read pdf. So I wrote a nice cover letter and with cover letter I attached my CV. So many people responded me back some even extended help and some some replied about that inability to help. So yeah that is how I know that this is at least at least this is working, this was working. And then later, when you collected all these emails these emails were of data scientists or HR people? Yes most most of the email was from HR and data scientists. So whatever the relevant email I thought uh I collected them and I used gmask gmask is like gmail extension that gives 50 email per day for free to send you. Okay. So I used that thing to send around 50 email per day like for I did this for like two or three weeks when when my email exhausted. I see. So so yeah this was like this was good so this was giving me some leads, some people were referring me some people uh were calling for interview so this kind of thing happened, and in later stage I collected I collected the contacts of lead data scientists over LinkedIn and I messaged them personally and yeah some of them replied me back. So yeah that is how I got some some leads for interview. All right so let's talk about your first interview experience. How did you get your first interview call and did that come through you cold cold emailing people? Oh yeah, that actually come from I called messaging people over LinkedIn. So yeah my first interview was on Editorialist yx so this is an american e-commerce company, so yeah that is a very nice company. So uh in the first stage they gave me some dataset and they asked me to share a jupyter notebook in one day after doing all of your analysis. And after this a personal interview followed and in personal interview mostly they they asked me question related to a statistics and some machine learning algorithms, and yeah some machine learning algorithms and some little very little python, not much python and after in second round uh a similar process happened in which they were they were asking me how to clean data, how to decide on clean data, how to impute data, when you should impute and when you shouldn't impute, uh what is CVSS score uh what can you get from the CVSS score. So these kind of things happen in the second interview. So overall it sounds like they didn't go cover python too much in detail. Yes that is right they were more focused on statistics and understanding basic understanding of machine learning algorithm. Do you remember any statistics question that you want to share here? Yeah of course so they they gave me a chart that was output of a stats model and he asked me uh what do you see there? So I saw p values were there and they he asked me what is the what is the meaning of p value just tell me what is the meaning of value p value, and what does it signify? So this is the one question I remember and uh some question was also there on Bayes' theorem because Bayes' theorem was Bayesian probability was written in my CV. So he asked me something about Bayes' theorem and yeah that is all I think I can remember some question also about the machine learning algorithm was there. For example uh he asked about various method of clustering for example dv scan, gaussian mixer and came in cluster so yeah these kind of things. Got it nice then you also received second job offer. So you got two job offers and then I think now you have two job offers and uh how do you decide like money money is one factor I understand, but then other than money like how did you decide like which job offer you want to accept? Yeah that is like that was a difficult part for me because both of the company was good and both were giving me like uh quite a good offer. I went for the another company uh The Math Company. I went there because the kind of work they were doing that were very much uh like fit for my profile my profile, they were doing research so yeah I was also doing research. So I thought that I would be good here. Nice so I really want to congratulate you on your thank you successful journey from astronomy physics to research data scientists which is so cool. If anyone is looking to transition from let's say who have non-computer science free uh degree and if they want to move to data science, what kind of advice would you give to that person? Uh yeah sure. So one basic thing I would like to say is uh technology like keep changing. So any extremely relevant topic, any extremely relevant technology now can go irrelevant today. So you should be focusing on like you should be learning how to learn and you should be trying to learn from the first principle as much as possible, as much as possible. And yeah of course learning technology is fine that will give you job, but in a long run you should be learning how to learn. So yeah one thing is that and uh and the second thing I would like to say, that uh if you are from physics or a stats or maths background so believe me you are required in the industry, believe me you just have to reach out. You have to reach out enough. So yeah there is still very much vacancy in my in market in industry. Uh but I don't know somehow right people are not reaching out to the right person. Yeah that is all I will think. Nice this is a golden piece of advice you have given which is learning how to learn and reaching out to people. I mean I love it. Yes and yeah third do a lot of projects or do like at least 50 or more projects and just present five and please and please don't present that California and EMNIST dataset because people are like getting bored of that data. Try to do yeah try to do something surprising. If you surprise your interviewer a little bit even a little bit that will go like that will be go with your side. Yeah this is what I would say. Surprise them with uh awesome project. Well thank you very much Shrish for your amazing conversation. We'll provide all the links in the video description below. And we'll all uh I is it okay if I provide your LinkedIn as well in the video description below? Yes sure sure. Sure sure yeah that that will be good. All right we'll do that. Thank you very much! Yeah thank you thank you Dhaval for having me. It was pleasure, It was nice talking with you bye! Yeah okay bye! [Music]
