Transcript for transcript_1IK-D1U6qQ-SSKIgo44iEM2XRnB22BaRR_1860_3000_cdd3522fe22e_3.txt

transcript_1IK-D1U6qQ-SSKIgo44iEM2XRnB22BaRR_1860_3000_cdd3522fe22e_3.txt

[00:00:00]  Sri Lakshmi, I think we can interview you.
[00:00:04]  Okay, sir.
[00:00:07]  Thank you, sir.
[00:00:08]  Thank you.
[00:00:09]  Thank you.
[00:00:10]  Thank you, Sri Lakshmi.
[00:00:17]  Can I ask the second student to join?
[00:00:19]  Sure.
[00:00:20]  Yes, please.
[00:00:45]  Hi, Likid.
[00:00:46]  Good morning.
[00:00:47]  Hi, ma'am.
[00:00:47]  Good morning.
[00:00:52]  Yeah.
[00:00:53]  Yeah.
[00:00:53]  Hello, Likit.
[00:00:54]  Could you introduce yourself?
[00:00:56]  Yes, sir.
[00:00:57]  Good morning to everyone.
[00:00:58]  My name is Likit Mamadi.
[00:00:59]  I'm a second year B.Tech student from computer science background at Chaitanya Deem to be University.
[00:01:04]  I'm currently working as web development intern at Carbolink Private Limited India.
[00:01:10]  So I joined them on 20th May.
[00:01:12]  Overall the 20 days I have worked with a rapid pace and I also built a complete corporate based waterproofing website from scratch to 85% and also fully stabilized their
[00:01:26]  sports category background also.
[00:01:29]  Overall, while I'm enjoying my full stack development, but my core passion lies in the generative AI side.
[00:01:35]  So to apply them, I have built practical application models such as RAG applications, smart AI documentation applications by using end-to-end RAG pipeline and also with help of
[00:01:48]  files and also with langchain and also built a built a fully full stack parent application where parents can get their child details from
[00:02:00]  that particular app of the child's data and i want a transition in my career background from a web development to ai sector so that's why i'm
[00:02:11]  I'm building this.
[00:02:13]  Okay.
[00:02:14]  So you have said in the beginning that you have prepared a website for
[00:02:19]  the current organization that you are working with.
[00:02:23]  Yes, sir.
[00:02:24]  Can we see that website?
[00:02:26]  Sure.
[00:03:03]  So here is the website that I have built for them.
[00:03:06]  Okay, so have you made it from scratch?
[00:03:08]  Yes, sir.
[00:03:09]  From the ideation and from even the design part also.
[00:03:14]  Like it's completely like up to 85% and here I have to like add the details and something and all the articles that they have made for the clients.
[00:03:24]  And the footer part was just and even the like about a section for systems regarding the products
[00:03:31]  and the product like they have like something like all of the products of waterproofing that they that they are doing so from here they can
[00:03:42]  get their complete system details of what they are doing and what their products were also capable of.
[00:03:48]  of and coming to the product section they have their complete waterproofing related products here so
[00:03:58]  the top products were placed at the top for like the most selling products of the waterproofing site and these are the documents
[00:04:08]  that of these products like how they're doing like that TDS and MSDS of the products and also have been done the other website also.
[00:04:29]  So this is the first website that I have done like from the 40th person to this production level ready because at this point they are mainly targeting waterproofing because it's coming the rainy season right.
[00:04:44]  So they are mostly on the side of the sports background.
[00:04:49]  In India they are dominating the sports background of their categories.
[00:04:54]  So the mainly income was and mainly targets were mostly sports.
[00:04:58]  So firstly we have built
[00:05:00]  and at present we have stopped that and we are doing the waterproofing website for the upcoming season so have you prepared made this website also from scratch
[00:05:10]  like from the 40th person because when i joined i have like i don't know that much regarding the website or something so they told me this is the base of the website
[00:05:21]  so from this the next part of the sub web pages like about us same like that is regarding the waterproofing it is it is completely on the sports side so there have multiple sector sites so for each sector we are building
[00:05:35]  a web pages like this and
[00:05:37]  And also I have done many things in their main website like Carbolink India also, like added many new features and now I'm presently doing on a
[00:05:49]  automation like where a user can get an automated quotation from the website or from the WhatsApp also.
[00:05:57]  Okay.
[00:05:58]  So, which the sales of their particular company will increase up to 30% there.
[00:06:03]  And how long have you been associated with Carbolink?
[00:06:08]  Like after my semester exams, like I have joined them on 28th September, 28th May 2026.
[00:06:15]  Okay, so today is 10th June.
[00:06:17]  So you prepared both of these websites from scratch within 20 days?
[00:06:23]  From 48th percent, the sports category have built and from the scratch until the 85th percent.
[00:06:30]  85% I have built the waterproofing website.
[00:06:33]  So you have almost made the entire waterproofing website by yourself in 20 days and you have prepared around 40% of the current website that we are seeing.
[00:06:42]  Yes, sir.
[00:06:43]  Okay.
[00:06:44]  Like I have to like before only I have some knowledge on the website designing part.
[00:06:49]  So.
[00:06:50]  So again, I took the previous works also of them, like the Carbolink India main website.
[00:06:56]  From that, I have took some designs and I worked with multiple web pages like designing parts.
[00:07:04]  So it made me simple to design the website.
[00:07:08]  Get ready that much fast and also have like day and night I used to work for that website because the season was rainy season is coming.
[00:07:17]  Okay, okay.
[00:07:19]  So Likit, I understand you have, you are interested in generative AI and more side towards the machine learning aspect of computer science.
[00:07:31]  Yes, like presently I'm building like generative applications for my better understandings.
[00:07:37]  But if you have read the JD that we have floated and we have provided to your organization, it only contains a very small amount of AI knowledge to have.
[00:07:48]  It basically requires the candidate to have knowledge about Python, JavaScript, Node.js, database and API configuration.
[00:08:00]  So do you have that knowledge with you, at least some working knowledge?
[00:08:04]  Like on the backend side, I have been completely worked on for six months.
[00:08:07]  I have built many applications also.
[00:08:09]  Before the Gen AI stack, I used to work on the completely on Golang.
[00:08:15]  They're completely backend for me.
[00:08:16]  I'm related with like all the APIs and all those things.
[00:08:20]  But for the Node.js part, I have to once go through the documentations and all those things.
[00:08:25]  Because the backend logic was same to all the languages, but only the syntax part and the other things.
[00:08:30]  Yeah, but I am asking more from an interest point of view because you have very early said that you are interested in Gen AI and ML.
[00:08:37]  I said that.
[00:08:38]  But as I said, the job requires, the job is not very much AI or ML oriented.
[00:08:45]  The current responsibility.
[00:08:47]  So that's why I'm asking, will you still be interested in pursuing this?
[00:08:51]  Like the stacks completely depends upon the Python node.js?
[00:08:55]  Yeah, no, no, that's fine.
[00:08:57]  What I'm saying is what the current job description or the key responsibilities of the role do not require a person with affinity or the expertise of AIML.
[00:09:10]  And because you are interested in AIML, even if you do, I understand you have knowledge about these things as well.
[00:09:15]  But like the job is not AIML oriented.
[00:09:19]  The responsibilities are not AIML oriented.
[00:09:22]  Would you still be interested in pursuing this?
[00:09:26]  Yes, sir.
[00:09:26]  But I would also love to work with the more...
[00:09:30]  challenging things like always on the AI.
[00:09:32]  The AI will make our things much more better, but making a production level products will also help me to tackle those challenges.
[00:09:42]  Okay.
[00:09:42]  I would love to.
[00:09:44]  Okay.
[00:09:45]  So the thing is that this job, it's not just a technical role.
[00:09:50]  It's also a coordination and managerial role.
[00:09:54]  So we want someone who can manage.
[00:09:57]  What do you understand about IoT systems?
[00:10:00]  Like IoT systems are also, in previous, like almost seven, eight months back, like I was a bit interested in IoT.
[00:10:08]  Even from the childhood side of childhood, I have built many like small, small IoT projects like by seeing in the YouTube.
[00:10:14]  But in our college, when I came to here, like I got to know around one and a half year after I'm joining the college.
[00:10:21]  So me and my friends thought to build a robot project like by using IoT and robotics.
[00:10:27]  So firstly, we have approached them and they gave a small task to build the IoT thing, like a self-balancing robot, where it can self on itself by calculating the angles, gyro and pitch yarn.
[00:10:42]  So how to balance a robot on a two wheels, I have learned by building that project.
[00:10:49]  So, like at starting of the point, it is very difficult me to understand all the fundamentals of a IoT like Arduino, ESP34 and all those things.
[00:11:00]  So, I took a one week of a break and I went through all of those things like the basic understanding of those things like the pins and all of those ESP32.
[00:11:12]  After I have done that project, even the coding part, even a small change in the code or even a small weight category also, it used to not balance the robot perfectly.
[00:11:24]  So those all things made me much more interested to work on the IoT more.
[00:11:28]  But at the end of the day,
[00:11:30]  In our college, the placements like regarding the internship companies, they used to ask me like the backend or the backend.
[00:11:38]  So I have to switch from IoT to backend and backend.
[00:11:46]  Okay.
[00:11:47]  So the thing is, in IoT, it's basically about communication and data transmission.
[00:11:54]  And the IoT is basically Internet of Things, where the things themselves talk to or communicate to the servers without any human intervention.
[00:12:04]  For this, we use SIM cards, we use cellular network, and we send transmit data through the devices to the servers.
[00:12:12]  Using the cellular network.
[00:12:14]  For that, we need someone who can manage and coordinate for the SIM cards that we use for those cellular applications.
[00:12:22]  So we generally use Jio and Airtel SIM cards.
[00:12:26]  And we want someone who can take ownership of this particular role.
[00:12:32]  Obviously, we will be there for guidance and if you have any questions to assist you there.
[00:12:38]  But we need someone who can take complete ownership of this role so that we can leave that responsibility with that person.
[00:12:47]  Yes, sure.
[00:12:48]  Sure, sir.
[00:12:48]  Like I would love to take the ownership of those things.
[00:12:53]  Which helps me to understand better things and also it will be helping me to gain more knowledge and experience on per
[00:13:00]  particular things.
[00:13:01]  Okay.
[00:13:02]  But as I said, but again, it's not something that you are very much interested in.
[00:13:07]  So I would like to confirm again, would you be interested in doing this?
[00:13:12]  I'm much more interested in IoT, sir, because we also like given a proposal like making an IoT model.
[00:13:19]  For railway, for student railways.
[00:13:21]  So whether like it got rejected in our college due to the funding issues and all those things.
[00:13:27]  So I have been doing the IoT things also apart from these things, but at present I'm not having that much time.
[00:13:35]  So that's why I'm much more interested in IoT rather than the full stack or the backend things.
[00:13:41]  Okay.
[00:13:42]  Is there any other project that you can show?
[00:13:44]  One more project?
[00:13:46]  Because you have done more projects till now.
[00:13:48]  Like on the JNA, sir?
[00:13:51]  Anything.
[00:13:52]  Anything that you think is like, would you, anything that you would like to show?
[00:13:57]  Sir, at present, like,
[00:13:59]  I have made those JNA projects like the basic chat like rag applications where the main goal of this project was to get
[00:14:13]  exact accurate data from the PDF or a file without any hallucination from the AI models.
[00:14:19]  So when we upload a multi-page applications or multi-page files, so
[00:14:26]  Some of the applications will get the data from the AI models.
[00:14:30]  even though the data was not present in the file.
[00:14:33]  So to overcome that thing, I have done this project.
[00:14:37]  So like, firstly, I will take the data from the file by reading the PY PDF loader.
[00:14:46]  So I will extract all the data.
[00:14:49]  And after extracting all the data, I will...
[00:14:52]  Pass through a function which is known as the recursive text splitter.
[00:15:00]  So in that I have set a certain instructions like the chunking part should be like 500 size and with the overlap of 50 characters.
[00:15:11]  So why have kept the 50 characters of the overlap?
[00:15:15]  Because while the semantic.
[00:15:18]  Semantic meaning of the sentence may not break in the half because if we keep the 500 characters over
[00:15:24]  chunk size, sometimes the character, like the model will take only the first 500 and the second 500.
[00:15:32]  But between the sentence, there will be a meaning regarding if we split the sentence into two parts, there will be no meaning so that we don't get the accurate result.
[00:15:40]  So after.
[00:15:41]  After getting the chunking the data from that from the pdf loader so i will pass through open source model hugging fish uh especially uh
[00:15:52]  llm v sorry lm v5 lm l6 v2 model so that it will create the 384 dimensional vector embeddings
[00:16:03]  So that the data will be stored in there.
[00:16:06]  And from that, I can store and index the values by using Firesys from the model from the Facebook.
[00:16:13]  So I can also get the clear, like fast and with zero latency.
[00:16:18]  And also it is good and capable for the small applications.
[00:16:22]  So when a user puts the query in the chatbot, like in the text box, the...
[00:16:30]  API will be fetching the query and it will also pass through the files.
[00:16:36]  So it will calculate like what are the related things to it.
[00:16:41]  So it will take the top three picks from the file.
[00:16:44]  Like if we are asking about what is meant by embeddings.
[00:16:47]  So it will take the top three picks about the embeddings and it will pass through it.
[00:16:52]  And also based upon the distance between the distance between the graphs like 384 dimensional it will pick the top three so those top
[00:17:04]  those top three picks and the query will be sent into a instruction instruction prompt that is given to a gemini gemini
[00:17:13]  API.
[00:17:13]  So on that API will be clarified.
[00:17:16]  We are given strictly that only give the answer which is present in the PDF or in the file.
[00:17:23]  Otherwise, don't give any extra answers by your own creativity so that it will only give the related answer to the user.
[00:17:30]  So that is the model that I have created.
[00:17:37]  So you have worked on RAG as well, right?
[00:17:40]  I said RAG.
[00:17:41]  Okay.
[00:17:44]  Fine, Likhit.
[00:17:45]  I think we can end the interview here.
[00:17:49]  So we will let you know what to decide.
[00:17:52]  Sure, sir.
[00:17:53]  Thank you.
[00:17:53]  Thank you.
[00:17:54]  Thanks.
[00:18:00]  Can I ask the third student to join?
[00:18:02]  Sure.
[00:18:20]  Hi, I'm Rita.
[00:18:20]  Good morning.
[00:18:24]  Hello.
[00:18:25]  Yes, ma'am.
[00:18:27]  We can start, maybe.
[00:18:29]  Okay.
[00:18:30]  Hello, Amrita.
[00:18:34]  Please introduce yourself.
[00:18:37]  Yes, sir.
[00:18:38]  Good morning.
[00:18:42]  pursuing BSc at BUS and also alongside with my degree, I'm also a student at NextWeb.
[00:18:48]  Recently, I have been exploring the field of AI, particularly areas such as generative AI, LLMs and RAG.
[00:18:57]  I have worked on a few projects involving