OH-04-07-2021.mp3-from OneDrive

Harpreet: [00:00:06] What's up, everybody? Welcome to the comet Emelle Open Office Hours powered by the artist, the Data Science. It is Sunday, July 4th. Happy Fourth of July. Happy Independence Day to all the Americans out there. I am American by nationality, born and raised in America. But I've been in Canada for the last several years, seven years or so. And we celebrated Canada Day just a couple of days ago on July 1st. So that's been that's been good. Had a super long weekend, been pretty much on vacation for like four days. This is great. Very, very awesome. We're happy to have all you guys here shout out to, uh, to Christophe and Meryn for joining in. If anybody else wants to join in, you can do so by following the link in the description of this chat. If you guys got questions on LinkedIn on Twitter, sorry, LinkedIn Twitch or YouTube, let me know right there. And I will be sure to get you guys in on the, uh, on the session. We'll answer whatever questions we can. Also, guys, uh, be sure to, um, check out the episode I released on my podcast earlier, uh, this week, as on Friday, actually, I did an episode with Dr. Jordan Ellenberg. He's the author of a book called Shape, which I have laying around here somewhere. I'm going to be given a copy of that book away. All you got to do is share this stream right here on LinkedIn and then you'll have an opportunity to get in and, you know, this free giveaway. But I'm super excited to have all you guys here. Cristoff Madden, good to see you again, Marion. Good to see you. Adam is in the building as well. If anybody else has, you know, anybody else want to join in? Well, wherever you are, there's a link right in the descriptions to come and join us. Cristoff, man, what you been up to?

Speaker2: [00:01:52] Sorry, I'm using my mobile right now for the first time, so I hope you can hear me. Yeah, I got married [00:02:00] so.

Harpreet: [00:02:02] Oh, you got married. I was just set up to say hello. Man. Oh man.

Speaker3: [00:02:06] That's all I'll say. I'm glad I didn't ask.

Harpreet: [00:02:11] That's cool, man.

Speaker2: [00:02:12] It was cool though. I took a week off and I'm. I'm back here right now. Nice.

Harpreet: [00:02:20] Welcome to the air. Welcome to the club, man. That's awesome. Thank you. Yeah. Man. So, man, I've been I've been really, uh. I've been thinking about what to study next and Data science. And one thing that I've been putting out for a while has been, uh, NLP natural language processing stuff. So I think partially inspired by, uh, Christof Hicks. I know you've been going in on on NLP and I know Meran was talking about going it on NLP. And I figured, dude, I sit down so much so much data for my transcript, for my podcast. There's so much text Data that I have. Um, I should do something with it. Right. I should see what I can work. I can get it going with that. So I've been doing a picked up this book called Applied Natural Language Processing for the Enterprise. It was a Riley book and it was a linguist, David Knickerbocker, who had shared it on LinkedIn. And I started going through it. I was like, oh, that's pretty good. Pretty good book. Um, so I enjoyed checking that out. So then slowly making my way through that book over the last week or so, I'm really enjoying that. What are you been using to, uh, to learn NLP Cristoff. So how did you start on your NLP journey. Um, I'm coming to you for advice on how to study this stuff

Speaker2: [00:03:28] And so on. My first thing was always this book. So this is where I started everything and this is what I like about NLP. And it was also the time when I was just trying to figure out what what I I'd like to do in machine learning. And I just found a couple of online classes only with me. And I went through two of them. One of them was about Tensorflow [00:04:00] 2.0 and the other one was advanced NLP. And from a guy called Lazy Programmer, I think I heard of him. Yeah. So that's how I started. And then I had some breaks. And in February this year I just discovered this spacy library and that's when I got. Got into it again, but I'm I'm still going through. I mean, I don't have, like resources, but I go to every time. It's like a lot of Google and reading and stuff.

Harpreet: [00:04:37] And so, like in the book, I'm using it. They're heavy on spacy and hugging the face. And so it's been cool and working with a lot of work with a pre trained model. How about taking the prepaid model, how to fine tune it using transformers and things like that? Because like do like I have so much text Data from from the podcast and do the transcriptions and all that. Um, there's got to be something I could do that's fun and interesting with all of that Data. So I'm excited to learn a bit more about NLP and get going on that, um, shout out to everybody else joining us. What's up Adam? What's up? Remote's up. Roger, if you have questions on anything whatsoever, please, by all means, go for it. And if you guys have questions that are of you on LinkedIn, on Twitch or on YouTube, let me know. I'm keeping an eye out on all of those chats.

Speaker4: [00:05:29] Yeah, hi. My name is Adam. I have a quick question. So, yeah. So I am in the process of working with a team here, doing a cargo competition for the covid x ray competition. And and so Niko from committable, I don't know if you know him, but he was gracious enough to give my team and I access to participate using that the tool [00:06:00] comment about for the Chicago competition for free. So that was awesome to of those guys. Right. So I'm going through the process of trying to use the tool along while trying to learn machine learning at the same time and keep learning. So specifically, we're trying to train these models to do object detection of the x ray and, you know, to determine that classifier between covid or no covid. And then in addition to that, being able to train the models to build and predict bounding boxes on the Data as well. So so this is all new to me. And I basically took this cargo competition to basically try to sharpen my machine learning skills. And so I was just curious of everybody. And if anybody out there has any words of wisdom of how to drive the common small tool in a way that you can optimize the hyper parameters, because now I'm in I'm getting into the realm of analyzing, you know, the model and then trying to train the model. And I'm just wondering if there's any some protests, if anybody's experience with that, how to approach optimization of the hyper parameters, basically. And I've done a lot of reading about it. But just curious what other people

Harpreet: [00:07:22] I think come has a lot of great tutorials available on their website as well. So, I mean, first of all, do that's a really interesting project, especially like as your first project for a machine that's like jumping off the deep end and it's crazy. But as long as you going to have fun and enjoy it, yes, by all means go for it. But yeah, it's tool is really cool. So just full disclaimer, I don't work at Comet or for comment, I just helping them out, building their community and doing stuff like that. Um, so they've got some pretty decent tutorials on their website that you can run to look at it. But essentially what it does, it allows you to manage and track [00:08:00] your experiments, um so you can look and see how different hyper of parameters affect your model metric. Um, and it does that for you really nicely. A clean visual kind of interface. Um, you just add like one or two lines of code and go for it. So they actually had a couple of had a couple of really interesting. Um, I actually had a computer vision tutorial on here. I'll see if I can dig my hands up on there and get it out to you, but um. Yeah. Like what specific question do you have about the parameter T I guess is, um, maybe you could start there. Well I think, I

Speaker4: [00:08:37] Think, you know, there's, there's different ways to do it. There's grid search. There's, uh, there's Bayes Bayesian optimization, there's random. And and I think what happens is when you start getting involved here, there's so many different parameters that you can manipulate to try to tweak the model. And sure, learning rate is a good example. You know, looking at their tutorials, you can modify that for the optimization parameter, you know, for the learning rate and momentum as well. But I'm just curious if other people had any experience with other parameters, because if you think about all the permutations of parameters you can sweep through as it goes, it's too crazy. Right. So. What is really the best bang for the buck? I was just curious if maybe there's other parameters to look out for. It's so general and every model's very specific. So it's not like one size fits all. But I just thought since it was for office hours, I could just throw this out there and make fun of anything.

Harpreet: [00:09:42] Yeah, definitely. So in terms of, like, search methods, I think definitely go for don't don't go for good search because I'll take forever. Right. Maybe randomize search Bayesian optimization. I think there's one called Hyper Opta as well. Those are, uh, tuning algorithms for high parameters. But in terms of which [00:10:00] specific high parameters should tune for object detection model, I'm not too well versed in that. Maybe. I think Meryn was looking into computer vision problems like that. Were you familiar with this?

Speaker3: [00:10:12] Uh, no, actually, remember, I had the first who they ventured into the division or NLP. And by the way, those NLP. Oh, nice, nice. But that wasn't just takes too much emphasis on Data is always very big and difficult. Well, from different sources though. Just because they felt that they should be part of the exercise of the faith based on the translation, we both like to think they will come up with no medical Data no not no experience, but the search optimization. But I mean, the four basic things is the message is love, Nicholas and baptizes think that both sides would be important. So no, no experience in so television models.

Harpreet: [00:11:13] Yeah. I appreciate that input there. Thank you. I think that's a general strategy. Adam, like what I would recommend if you I mean, this is a great opportunity for you to take that individual algorithm that you're looking at and really try to explore it and get a deeper understanding of it is just tune one parameter at a time, holding everything else constant to whatever the default about these are, and just see the impact that tuning that one parameter has on your model outcome. Right. And so comment MLS's tool really allows you to do that because of how like visual everything is and how clean everything is laid out. So you might want to take this opportunity to explore whichever algorithm you're looking at and go deeper thinking that that intuition behind how that individual hyper parameter, uh, affects [00:12:00] your model metric. That's great. And that's what I would do. I've not done any, like, object detection or anything like that before, so I'm not too well versed in that. Yeah, that would be the approach I would take. Yeah. Yeah.

Speaker4: [00:12:12] That makes sense. That makes sense. Great. Thank you for your feedback. Yes.

Speaker3: [00:12:18] Just a quick question. I was thinking prebuilt basically and then something like business or something. Yes.

Speaker4: [00:12:26] Yeah. So, so ok. So yeah. So we're initially OK so the goal is to try to take advantage of that YOLO. OK. Right. That's for the the pounding box object detection mechanism or the fact that a section of the contest, the other thing that we're trying to do is build like inside YOLO, you, you insert a backbone and you select which model you want to start with. And like you said dence, that was one of with that.

Speaker3: [00:12:59] Yeah. I'm going to go about turning the computer with the most basic.

Speaker4: [00:13:08] Yes. Yes. So so I'm still learning about the details of how that work under the under the hood and again and those actually they have their own parameters that you can add to it too. Right. So, yeah.

Harpreet: [00:13:21] So I'm going to point you to this reference here. I'll put it up on the screen here. So these are some of the tutorials and walk through that comment has regarding in this case how to debug object detection models. Um, I don't know if you've seen these yet, but these might be really helpful for at least getting started to get an understanding for where you're coming from. OK, well, go ahead and I'll link this in the chat. Thank you.

Speaker4: [00:13:43] Yes, I think the topic I may have missed this. I've been scavenging their site, but yes, I, I think I may have overlooked this one. But thank you for pointing this out.

Harpreet: [00:13:52] There's another one here as well. A link to its team comment, metal object detection there. Looks like this is a notebook [00:14:00] that you can look through. So go ahead and put that here for you as well for a couple of things to to get your answers to. That can be more helpful for you beyond that. But definitely play out the tool, explore it, gain that intuition about each one of those type of parameters by tweaking one at a time and see what happens. But, um, they should help you out to to really understand how to use comit for that particular purpose. Yeah. Thank you also. Well, good luck. I'm excited to see how this turns out.

Speaker4: [00:14:27] Freeman Yeah. Well, you know, we have about another month ago or one month end. And so we'll see how it goes.

Harpreet: [00:14:34] Right. It looking forward to, uh, to to see how this works out. All right. All right. Uh, shout out to everybody else. We got Rakesh and Rhema Barens in the building as well. Um, looks like there's some questions funneling in through LinkedIn. Um, yeah. So is joining a bootcamp beneficial? Because it's a lot of money I do for self learning process. It depends. Are you the type of person who can effectively learn by yourself if you need more structure and more rigor and you need something that's laid out for you completely, then looks like you can do self learning on your own. But if you need that guidance, if you need that instruction, if you're the type of person that needs that, then the boot camp might be beneficial. And there's always that aspect of this. It's when you invest money into something, you feel more committed to it. You feel like, OK, well, I've paid money, so I should follow through with this. Uh, so there's that benefit to it depends on how you want to, um, you know, look at that. But, um, depends man. Depends on on you. Uh, so, yeah, that was I don't know if that's the same. That was a case that was in the chat asking on LinkedIn you should just ask him and

Speaker3: [00:15:43] Go for a man I have experience with. But I started to online courses and maybe it's my personality. But again, online courses for me at end of the unstructured and the value can be very messy. I have taken maybe ten [00:16:00] classes on different subjects when I started. I mean, basic introduction to this machine learning both of them, the initial three, four and Cytron different. The problem is that if you're new to that, the science, the field is so humongous that it's very easy to get lost. And after one year trying to learn on my own, I decided to join a bootcamp. There are several camps like that. This agreement for payment. After you get the job, you can sign every I forgot what the name was, but basically I have to get the job. You pay for the boot camp thing. But I attended the one that stands boot camp and it has the same agreement, the same problem. It's original is six months, but there is so much material that they're severely underestimated the time that it takes to get through the critical mass of nine months. They had taken the extension and I'll tell midlist it made a big difference because they just like to sit down. And I would think because we cannot learn every preconceive math problem. But this place, it has a logical flaw is that at some point and then it's more evolved and complex subject. And that's helped me. I can do to tell the difference between that I did a year ago at my final conference. So my vote goes for both camps because this stacked up and you'll find a good boot camp that we can actually have pay off to get rid of that.

Harpreet: [00:17:40] Thank you very much. So, Rakesh, you're in the chat right now, so go ahead and let me know what you're saying, which I think in.

Speaker2: [00:17:51] Well, I found the causes and in can country like a replacement as the soldier or Marine, so. I thought about the [00:18:00] Data is great, but it feels like I'm kind of because because the school's curriculum is just for national and learning so much as it is, it's kind of hard for us. But since I wasn't in college there, it's been three years now. And still we're learning a lot of stories in this. And they're saying like it contains a lot of technical and industrial production projects. So it's still beneficial for us.

Harpreet: [00:18:36] I mean, it depends it depends on what it is you're trying to do. What's your what's your end objective? Like what what is it that what is it you hope to achieve by going to boot camp?

Speaker2: [00:18:46] Well, I was just going to use all this time that I had learned in my college education and training these boot camps. So I was trying out under this to try to learn all this thing on my own. But I found it very difficult and time consuming desert. Effective to do that,

Harpreet: [00:19:13] I mean, again, like I said, depends on your personality. For example, me, I could self learn anything like, I don't know, I could just find resources, create a learning plan for myself and just go through it. Right. Um, some people are like that. Some people need somebody to like, say, do this and do this, then do this, then this, this and this industry. So that is you. Then it might be beneficial for you to enroll into a boot camp if you have no discipline to keep yourself accountable, to hold yourself accountable, to create a learning plan for yourself. Right. Um, otherwise, everything's out there and everything's open source and Data science, like all the knowledge you need is out there. You just need to be able to set a plan for yourself. So, again, this is highly subjective. Right. So in terms of beneficial, that benefits are going to vary per person. I I'm that's I'm like asking you what your end objective [00:20:00] is like. What do you want to accomplish by going through a boot camp. Right. If you're hoping that by going through a boot camp, all of a sudden this indicates that I'm hirable, that I should be able to get a job, that's probably not the right rationale because you can spend that time in a boot camp by just building a project for yourself. Right. So what is your end objective that you're trying to to do?

Speaker2: [00:20:22] I just don't get hired and in some multinational company.

Harpreet: [00:20:27] But then do some projects, right, do some really well thought out projects that it could be any project that you are interested in. There's not like a magic Data set out there that's like, oh, my God, if I see this Data said, if I work on this, I'm going to get. No, that doesn't exist. Hiring managers don't care about what Data said that you are working with. They care about the way that you approach a problem, right. The quality of your code, the quality of your solution, the quality of your communication of each step and why you're doing it step right. So the biggest thing you could do, biggest time investment return on time investment will be doing a project for yourself because you're learning the skills and you're going to be applying the skills and you're building a portfolio. So now you're you know, it's just like a three fold a multifold, a benefit to that. Um. So, I mean, think about what it is that you ultimately want to do and.

Speaker2: [00:21:24] Yeah, but it is pretty much so. It's really hard for us to find a good cause for a good platform for us. I was trying my head with that code a while back, but

Harpreet: [00:21:38] So like you're already done with school. What did you study in school.

Speaker2: [00:21:43] I started doing statistics.

Harpreet: [00:21:45] Ok, well you got the baseline fundamentals, you know, enough math. So anything you can pick up from there, you just build on top of that and, you know, go for it. Right. Like, there's not going to be a magic boot camp. They just teach you how to be the best data scientist in the world. That just doesn't exist. Right. [00:22:00] So quit chasing the bootcamps. Just something is going to be my biggest advice. So the time that you spend in a boot camp, just take that time redirected towards an actual project, stumble, learn. And, you know, growing pains through that project is what you want. Right. So hopefully that is helpful. We're going to go ahead and move on from this particular topic. Does anybody else have questions on anything, uh, shout out Tanisha's in the building or subprime stuff?

Speaker5: [00:22:32] Yeah. Yeah, I can. I can go. Yeah. Go for it. Yeah. Thank you Harp for hosting the office hours. I really appreciate the chance to get to collaborate and and kind of talk a little bit about what I'm working on as well and get some feedback from the community. And so currently right now I'm working on an Agritech project right now with the international team. We're utilizing so satellite data and we're utilizing some sensor data from under the ground as well. And this particular project, it's a not for profit project. It's kind of closely linked, politically linked to the farmers protests and kind of helping out in that case. And more so what this technology does, it optimizes and saves water, especially for drought affected and developing countries like India, like, you know, it can be held for many different climates. So there's there's a lot of research and development that's being done right now. There's a prototype that's set up in Hunan, Maharastra, where there is an AB test that we're doing where half of the farm we're using this technology and the other half is the manual farming. And we're seeing, you know, they're they're optimizing the water by by more than half. Also, the crop yield is growing two times more so. And it's an automated the automated irrigation [00:24:00] system, drip irrigation system, basically. And so we want to scale this to the West as well, because as you grow, I don't know if you guys have heard, but there's a huge heat wave that's happening in California, in B.C.

Speaker5: [00:24:13] and those areas are heavy for agriculture as well. So we wanted to reach out to people who are in the industry as well and collect some like market research. So there's like this Google farm that I'm trying to get around. And I know that you're also from Manitoba and around the prairies. I believe that it's also a drought affected area. I don't know what agriculture is like there, but I'm just thinking, um, does anyone have any ideas on agriculture, agritech industry and and right now, kind of what we're working on while I am in the process of creating a prediction algorithm, using ASTM to basically predict when derogate next, using all these different data sources, mainly the sensor data from under the ground, we're actually entering a competition, an Eastern European competition. And sometime next, I think later this month, the deadline. So it's called Apricus Masters trying to get some investors and funding as well. So it's kind of, you know, a lot of moving pieces. And I'm just curious if anyone has any feedback or questions or anything that can help out with this that is known for. So, yeah,

Harpreet: [00:25:30] That, like, precision precision agriculture is huge. I think that's such an interesting application of machine learning. There's a company that's based out of Winnipeg called Farmers Edge. Are you familiar with them at all? Have you heard of them? And they do exactly what you're talking about. And they've got like a whole squad of their scientists and stuff as well. And they're doing exactly the same type of concept where they're looking at satellite imagery data and then they have sensors for weather around different parts of the of the the plot of land. And they're just predicting, OK, which [00:26:00] area do we need to provide more like irrigation or nutrients or whatever to so super fascinating. I think that's a really, really interesting idea. I would just look for what recompetition would be doing. Right. So just look at some of the companies that are already in the space,

Speaker5: [00:26:16] The farmers edge. Yeah. The thing is, with this technology, like most of these technologies are really expensive. It's not accessible to small farmers. And that's the issue. A lot of the smaller farmers in these developing countries, they don't have access and their water is a precious resource that's going to be running out. You know, climate change and global warming is something that's huge. And no one's really I don't really see a lot of people doing anything about it, and especially in developing type of countries. So this product itself, if we got it at an extremely low price point, so it is affordable. That is like the main purpose of this. And we're trying to make this affordable and reliable as well. So that reliability factor, that's what we're really working on and on the precision of the algorithms and the sensors as well. And we took a look. We did like a test on the satellite data and the sensor data and try to put the soil temperature on the soil moisture. We looked at if there were any discrepancies from the sensors on the ground and in the satellites and it lined it aligned. So, so far, so good like that with reliability. And democracy is pretty good. And we're taking things one step at a time. And it's and it's getting some recognition. But I just wanted also to see if. Yeah. So Farmers Edge, that's one company that you would like us to look into.

Harpreet: [00:27:38] Yeah. They've got them. So farmers that just like that, they just recently went public on the Canadian Stock Exchange. So definitely check them out. And then there's this actually there's an organization here in Manitoba that's based out of Manitoba is called the Enterprise Initiative for Machine Learning and Intelligence. And they are heavily focused on [00:28:00] applying machine learning to farming. And so you can look at all the partners. They have lined up with that, too. And this is a nonprofit board as well. But just take a look at what some of these people are doing. Right, because they they'll talk about stuff on their blog post or some of their data scientists might have written written blog posts or articles that are like tutorial type. Right. And just kind of see what it is that that they're working on and let that kind of inspire what you're doing. But I think it's like super interesting. Definitely. Definitely super interesting and a good cause. Asha has a question here. She wants to know who is the target market.

Speaker5: [00:28:39] Yeah, the target market would be small farmers and, you know, it could be big farmers. But right now, from what we've seen, a lot of farmers use timer based irrigation systems, not so much like, you know, automated and drip irrigation. And so they still waste a lot of water even though the farmers in the West. Yeah. So kind of our target market is small farmers to start collecting in our database right now. So that way, as we have released information, we can do some email marketing. And so that's kind of how we're going so far and just trying to reach out to as many small farmers as possible because they think they will be the most benefit from this technology. And the price point is.

Harpreet: [00:29:19] Well, yeah, I mean, it's better to like if you have limited resources, you don't want to just uniformly use resources on one Putland. You want to use it strategically on plotts pieces of that land that need whatever resources. Right. So rather than just like going through and spraying, let's say, fertilizer uniformly across the entire plot of land that you have, just focus on the areas that need it. You can save yourself money, can save resources. I think it's awesome. Great, great idea. And we pull this up real quick just to show you what's going on here on my screen. This is the website for Enterprise Machine [00:30:00] Intelligence and Learning Initiative. That's what it's called, EMTALA. And these guys are all about just using machine learning for agriculture.

Speaker5: [00:30:10] So we're looking at the drought affected areas in Canada, Manitoba and Alberta. They came out pretty drought affected. So I really have connected with you because I thought that we would have this info.

Harpreet: [00:30:22] Yeah, yeah. I think this is super cool. I would love to work on, like, a, uh. It doesn't like that because that's just fascinating, think you're using, like satellite imagery, Data, sensor, Data, uh, all sorts of stuff and using it for good cause. I think I think Mary might have had a question here for you or a comment or some go for it.

Speaker3: [00:30:41] Actually, I had a question, but they showed it on the screen about the length of the organization. And this was m i l l i yeah,

Harpreet: [00:30:50] Yeah, yeah. I'll put a link right here on the chat as well. Yeah. That's a great, great idea. Great initiative. Um, definitely look into I guess some keywords like uh would be like precision agriculture type of organizations and things like that. And I mean I would even see if I could find some like projects that people have done. Let's see if we can find something real quick. All right. Let's see if we do if we can find any Jupiter notebooks. All right. Well, here's a couple of things that you can look at. Propertied semantic segmentation. That's interesting stuff. Uh, just in case anybody wants to see what this would look like in action, uh, I'll put a link right here in the chat and then also showed a link on LinkedIn. Um, yeah. I mean, that's that's cool, man. I think you're doing a great thing with that. And look at this crop wheat field image. Data said, you see, other people have been doing, um, so linked to both of these in, uh, in the chat here. Thank you. Yeah. Awesome. Or any other questions or comments for, uh, Rhema here on her awesome initiative and go for it now.

Speaker3: [00:31:55] Somersaulted in a session. If I [00:32:00] had a question about this on my mind right now, does anybody in the meetup PDF experience that into the real deep into the circle of data scientists?

Harpreet: [00:32:14] Yeah, it's pretty standard part of the process.

Speaker3: [00:32:18] So sodium Asadi If you think you need more time, I'll give it to you.

Harpreet: [00:32:25] But I check out to check out. These are a couple of links here. Um, I mean it might just be interesting just to see what other people are doing so you can use that to kind of inspire what you're up to. But, uh, great, great idea. A great project. I'm looking forward to seeing how this, uh, pans out.

Speaker5: [00:32:41] Yeah, thanks. I feel free to reach out to me if you have any more questions or suggestions or anything interesting around the topic. And I'd be more than happy to chat.

Harpreet: [00:32:51] Awesome. Yeah. Sounds good. All right, Marianne, go for it.

Speaker3: [00:32:55] I just ask because there is some sort of confusion when you apply for a data center, this user that is causing role in the law companies. Does somebody experience with what kind of questions from the audience? Is it the typical stuff done to stop the that comes and so on or something particular like to manipulate the data?

Harpreet: [00:33:27] So in my experience for data science job interviews, the coding interviews I've gotten have all been problem solving type of coding interviews, not like, um, software engineering type of interviews where you're trying to do, like, you know, binary tree or something like that and, you know, LinkedIn stuff like that. It's for me it's it's been mostly, um, like problem solving type of questions which which you can find in like Harpreet Sahota or code hacker Reinke or Lete code, um, those type of questions. I mean one the one thing that I really enjoyed by is said this website here to kind of help [00:34:00] get an idea of what problem solving type of questions you might get ask is Python principles. And they do this. And this is great just for like, you know, building a little bit of confidence with respect to doing a coding challenges. They've got these challenges here, our python principles, dot com for such challenge. And they got like all these interesting, interesting types of, um, uh, challenges. They're fun. They're fun to do. Another good one is that I think stir, uh, things stir on YouTube and on YouTube.

Harpreet: [00:34:31] He's got, uh, these really cool, um, series of videos, uh, which I'll pull up as soon as this thing loads. And right here in the coding interview. And these are really fun and interesting ones. Um, so things like this like find the Keith largest element in and unsorted less things like that. Um, yeah. So more problem solving types of questions and they're generally pretty fun to do. They're interesting to do. I'll drop a link right here into the chat for that. Um that right there is the link to uh links to his playlist on um coding interview questions. And the other one was Python Principles. So yeah. Yeah, they're there in my experience. I'd love to hear anybody else, uh, anybody else's experience. But for me they have mostly been just like, uh, just Problem-Solving. OK, here's a. Let me think through it and let me think through the solution, um, and and how to get there with them programmatically, uh, Rhema or Nesha or ISIL or Cristoff and you guys interview processes. What types of questions have you guys come up with me?

Speaker6: [00:35:43] I think I also have had the business case scenario questions, so sort of a complete business. Usually at the end stage of an interview, the first couple of them are actually problem solving more more computer science oriented [00:36:00] coding questions. There are some some companies that do several levels of coding interviews. The person that used to be is probably Taringa. But I have had some companies that do also business case that that sometimes I just feel like they're giving a problem that they're trying to sort through the well, through these candidates. Yeah, because those sound like bad. Those are problems that they're trying to solve because it's facing that website, too. You just don't know whether they've already solved it or not. You're trying to brainstorm. Kind of.

Speaker3: [00:36:35] Yeah, yeah. I always that's a good thing for the company. That's my problem on their website. And see the following some problems.

Speaker5: [00:36:47] Remgro for just doing projects in general and showcasing your projects and nowadays it's always on so you can share your screen and then just show like what kind of projects you worked on that having visuals is really good for interviews. And in my case,

Speaker6: [00:37:03] One more thing one more thing I've had before from one of the companies is they showed the actual SQL and ask us to explain as to what might be going on. Maybe this is just in the health care domain. I only applied to get domain jobs, but they should because they're looking for a specific expertize, like SQL or a procedural language kind of thing. They just show it and ask you to explain what is going on in there. And sometimes they even ask you what the function does. And if you have really worked at SQL, you would know it's not something that that's something I've experienced just.

Harpreet: [00:37:43] But yeah, it's actually brings me to an interesting point. Right, because so there's kind of different levels of the interview, obviously, versus the iPhone screen, but they'll have like a tech screen itself, which might just be ten questions just to just to screen people out. [00:38:00] But then even after and that one of those ten questions could be like, you know, coding challenge type of thing where you're trying to problem solve using programing. Right. And then further along down the process, you'll get the take home challenge. I think the take home challenge is much more aligned to it. Nicias, talking about what I was talking about as well. So you don't get challenged just once you have like that screening assessment and then you go into the proper take home assignment. Does that help clarify that? And also, there's a comment here from Asia. Asia's joining us. She's on the road, but she's chatting right here. And she said that, uh, hers have mostly been code questions, but she's also received business case projects to, um. But a lot of the times it's been timecode challenges. And that's in my experience as well. You have a time to coding challenge. That's the kind of screening phase and it will get a take home assignment after that. So typically go our phone screen tech screen interview with the hiring manager or senior member on the team take home challenge and then another interview or two interviews after that.

Speaker3: [00:39:05] Thanks. I appreciate the help. Yeah, definitely.

Harpreet: [00:39:08] Yeah. I mean, best way to prepare for all that stuff is just do as many practice problems as possible. So go through those couple of links that I shared with you, like the fixture academy link with all those coding interview questions and then type on principles.