Nishant Kumar

𝗣𝗮𝘀𝘀𝗲𝗱 2 NPTEL 𝗲𝘅𝗮𝗺𝘀 𝗶𝗻 𝗷𝘂𝘀𝘁 2 𝗱𝗮𝘆𝘀! 🤯 Despite a hectic schedule and limited preparation time, I managed to pull through and achieve respectable scores in 𝗗𝗕𝗠𝗦 (68) 𝗮𝗻𝗱 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 (72). H͟o͟w͟ ͟d͟i͟d͟ ͟I͟ ͟d͟o͟ ͟i͟t͟? • 𝗙𝗿𝗶𝗲𝗻𝗱 𝗣𝗼𝘄𝗲𝗿: My friends ,Shashank Kotagi & Manish S M, were instrumental in helping me grasp the concepts. • 𝗖𝗿𝗮𝘀𝗵 𝗖𝗼𝘂𝗿𝘀𝗲: I focused on assignments, watched problem-solving sessions at 3x speed, and used other resources 𝗘𝘅𝗮𝗺 𝗗𝗮𝘆 𝗔𝗱𝘃𝗲𝗻𝘁𝘂𝗿𝗲𝘀: Traveling 25 km to the exam center was quite the adventure! But the exam itself was manageable, similar to JEE Mains. Most were MCQ so finished it in under 1.5 hours. 𝗔 𝗕𝗶𝘁𝘁𝗲𝗿𝘀𝘄𝗲𝗲𝘁 𝗠𝗼𝗺𝗲𝗻𝘁: While I missed the silver certificate by a narrow margin, I'm still proud of my achievement. A big thanks to Prof. Engr. Rakesh Kalshetty for recommending these courses. 𝗧𝗶𝗽𝘀 𝗳𝗼𝗿 𝗙𝘂𝘁𝘂𝗿𝗲 𝗘𝘅𝗮𝗺 𝗧𝗮𝗸𝗲𝗿𝘀: 1. Complete assignments on time. 2. Watch lectures at 2x speed. 3. Practice with problem-solving sessions. 4. Revise assignment questions regularly, most of them come in exam 5. With proper focus, you can aim for a gold certificate even with a week's preparation. The 𝗡𝗣𝗧𝗘𝗟 certificate is valuable, so don't miss this opportunity. It also offers 𝘁𝗿𝗮𝗻𝘀𝗳𝗲𝗿𝗮𝗯𝗹𝗲 𝗰𝗿𝗲𝗱𝗶𝘁𝘀. All the best to future exam takers! 🎉 #NPTEL #ExamSuccess #AgainstAllOdds #LifelongLearning #IIT

34

6 Comments

Hello all!! I'm happy to announce the completion of the Python for Data Science course from NPTEL taught by Prof. Raghunathan Rengasamy from Indian Institute of Technology, Madras and I scored 69/100. This course covers fundamental concepts of Python programming, including data manipulation with libraries such as NumPy and Pandas, data visualisation using Matplotlib and Seaborn, and practical applications in regression and classification, all applied to real-world datasets. NPTEL IIT Madras #python #datascience #pythonfordatascience #NPTEL #IIT #IITMadras

34

3 Comments

Please refer my previous post here regarding a blog which I have published in Analytics Vidya (https://lnkd.in/dhnJ-NfD) wherein I have detailed application of AutoEncoders in Cricket using PyTorch. In continuation of this, please find overall statistics of Top 2 Indian players identified by AutoEncoder as Anomaly. Train Dataset comprises of players from Test playing Nations other than India. Test Dataset comprises of Players from India. Dataset and complete code of AutoEncoder is available in Kaggle as under: https://lnkd.in/gN84WkWq What are your thoughts on these below statistics?

17

2 Comments

Indian Govt free internships and courses. Ministry of Skill Development and Entrepreneurship, India and Government of India Official Ministry of Youth Affairs & Sports, Government of India NITI Aayog Official Dharmendra Pradhan Govt courses and Websites that helps you upskill and battle Unemployment. Unemployment in India is 9.2% as on June 2024. And most of these in the IT are structural unemployment wherein the skills doesn't match with the market's demand. And a lot of us go with blame the government for not taking any actions. But the Government has taken a lot of initiatives to up skill students and even working professionals. But it's all little known. I got to know of it recently, but better late than never. Sharing it with you all so that we all learn together. Here are a few websites : 1. NPETL https://nptel.ac.in 2. Skill India Digital Hub Courses https://lnkd.in/gqDwAwGr 3. National Skill Dev https://lnkd.in/gbQYrQjF 4. AICTE skills dev Internships https://lnkd.in/gBj7bH2q 5. NATS Portal https://lnkd.in/gbyYaChA 6. For Electrical and Electronics https://lnkd.in/gCWB2FDN. 7. Skill Development and Entrepreneurship https://www.msde.gov.in/ 8. ISRO-IIRS e-learning free courses https://lnkd.in/gDcvEz2j Skills Acquisition and Knowledge Awareness for Livelihood Promotion (SANKALP) and Skill Strengthening for Industrial Value Enhancement (STRIVE) to boost the Skill India Mission. Both are central sector with half of the scheme outlay as loan assistance from the World Bank. There are a lot more provided by International websites, institutions and fora too. Follow Niranjan Sivakumar for more such updates. #hiring #hire #live #coding #skills #linkedin #jobs #india #leadership #careergrowth #innovation #career #management #AI #entrepreneur #business #GoI #sustainability #internship #intern #techtrends

5

Hello Everyone! 👋 I'm thrilled to share my latest blog post on "Language Modeling and Its Applications" based on my internship project at Innomatics Research Labs. 🤖 In this blog, I explore: What is Language Modeling? Applications of Language Modeling in NLP Steps to Model a Language 🔗 Dive deeper into the world of language modeling and its significance in NLP by reading my blog post. #genai #languagemodelling #llms #ai #nlp

27

My First Phone Interview Experience with a FAANG company as an International Student in US🙂 𝐑𝐨𝐥𝐞 : Data Engineer 𝐃𝐮𝐫𝐚𝐭𝐢𝐨𝐧: 1 hr  The interviewer introduced himself and made sure I felt comfortable, even encouraging me to take a moment if I needed water :)   Later he proceeded with below questions.  1. What complex data engineering task have you encountered, and how did you solve it? 2. What specific tools have you used in your ETL process? 3. How much volume of data have you handled? --> Now he asked questions about any projects that I did related to Data engineering which I said Yes. 1. Give a high-level overview of the architecture used in my project and the tools I employed. 2. How was I able to fetch the data(API or Files) in project and how did I transform it and how did I perform dimensional modeling? 3. How have I been able to conclude Facts and Dimensions and Why did I choose one schema over another?(Deep discussion on Dimensional Modeling for a quite a bit of time focusing on my project) 4. How much volume of data I have handled in the project ? --> Now some more questions on dimensional modeling, OLTP vs OLAP 1. What is dimensional modeling what is fact and what is dimension? 2. What schema have you used in your project, what is the reason behind you choosing that schema? 3. STAR vs Snowflake 4. Which Schema is preferred in what cases and Why? 5. What is OLTP and OLAP? What are the differences between them? What factors do we need to consider while designing OLTP vs OLAP? How would you do Data Modeling? 6. Laws of Normalization? Which Normalization is preferred for OLTP and why? --> Some questions on PySpark: 1. What basically happens when Spark application runs 2. Transformations vs Actions 3. Narrow Transformations vs Wide Transformations This discussion wrapped up in about 30 minutes. The remaining 30 minutes were focused on coding challenges in Python (We can choose any language) and SQL. 𝐒𝐐𝐋 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧: 𝐓𝐚𝐛𝐥𝐞 𝟏: 𝐅𝐫𝐢𝐞𝐧𝐝𝐬  Id    Friend_id 𝐓𝐚𝐛𝐥𝐞 𝟐: 𝐑𝐚𝐭𝐢𝐧𝐠𝐬 Id    Rating 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧: Retrieve all the Ids of a person whose rating is greater than their friend's rating. If a person doesn’t have any friends, retrieve their Id only if their rating is greater than 7. 𝐏𝐲𝐭𝐡𝐨𝐧 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧: Valid Parentheses I thoroughly enjoyed the interview and learned a lot from it (Credit to my Interviewer). Although I wasn't selected to move forward, I’m determined to keep trying and applying because that’s what truly matters to me. I encourage my fellow international students to do the same. I’m sharing this experience in the hope that it might help at least one person 🙂 . #Faang #PhoneInterview #DataEngineer #SoftwareEngineer #LifeAsAnInternationalStudent

854

29 Comments

Pleased to share my recent publications: 1. Beniwal, R., & Saraswat, P. (2024). A hybrid BERT-CPSO model for multi-class depression detection using pure hindi and hinglish multimodal data on social media. Computers and Electrical Engineering, 120, 109786. Elsevier. [SCI Indexed IF: 4] https://lnkd.in/gFYEuaEc [Link to freely download the article] 2. Kumar, N., & Beniwal, R. (2024). A Multi‐Constrained Green Routing Protocol for IoT‐Based Software‐Defined WSN. Concurrency and Computation: Practice and Experience, e8306. Wiley. [SCI Indexed IF: 1.5] https://lnkd.in/grr2fHw6

65

3 Comments

After completing nearly 5 years in compiler industry finally feels like home coming after joining IIT Roorkee as an assistant professor.

103

7 Comments

I tried openA1 o 1 today my favorite IIT-JEE problem. The question is an integral which has a very interesting answer :22/7-pi it one of the best IIT JEE mathematics problem which apeared in IIT_JEE 2010 Here are the result of when i asked this question to openA1 o1 abd GPT-4o openA1 o1:Gave wrong answer. GPT-4o:gave wron answer iexpacted openA1o1 to be good at advanced mathematic problems

1

🚀 Diving into ML: Day 6 - Support Vector Machines (SVM): Finding the Perfect Boundary Hello connections! Today, we're exploring Support Vector Machines (SVM), a powerful algorithm used for both classification and regression tasks. 🧠 What is SVM? SVM is a supervised learning algorithm that aims to find the optimal hyperplane that best separates different classes in the feature space. It's particularly effective for high-dimensional data. 📊 Theory: Optimal Hyperplane: Maximizes the margin between classes Support Vectors: Data points closest to the hyperplane Kernel Trick: Transforms data into higher dimensions for better separation Key Concepts: Margin: Distance between hyperplane and nearest data points Soft Margin: Allows some misclassification for better generalization Kernel Functions: Linear, Polynomial, Radial Basis Function (RBF), Sigmoid 🔍 Considerations and Challenges: Choosing the Right Kernel: Different kernels suit different data distributions Trade-off between model complexity and generalization Curse of Dimensionality: SVM can handle high-dimensional data, but performance may degrade Feature selection or dimensionality reduction might still be beneficial Scalability: Can be computationally expensive for large datasets Specialized implementations like SMO (Sequential Minimal Optimization) help Imbalanced Data: SVM can be sensitive to class imbalance Techniques like class weighting or SMOTE can help Hyperparameter Tuning: C (regularization parameter) and kernel-specific parameters need tuning Grid search or random search with cross-validation is often used 💡 Advantages: Effective in high-dimensional spaces Memory efficient (uses only a subset of training points) Versatile (different kernel functions for various decision boundaries) Works well when classes are separable 🔧 Tips for Optimization: Start with linear kernel for interpretability Use RBF kernel for complex, non-linear boundaries Normalize features for consistent influence Perform thorough cross-validation for hyperparameter tuning Consider ensemble methods for further performance boost 🌟 Applications: Text classification Image classification Bioinformatics (protein classification) Handwriting recognition Financial analysis (e.g., credit scoring) SVM's ability to handle complex, high-dimensional data makes it a go-to algorithm for many machine learning practitioners. Its elegant mathematical foundation provides both power and interpretability. Next time, we'll explore another fundamental algorithm in machine learning! #MachineLearning #AI #DataScience #SVM #SupportVectorMachines

16

As a visually impaired student, AI (mainly ChatGPT) has helped me a lot with my academic learning. It gives me solutions for many problems. For instance, when I was preparing for the JEE examination, due to a lack of accessible resources, I found it very difficult to study the tough concepts of PCM. But ChatGPT made it a bit more manageable to understand the concepts. I used to prompt the problem and ask for a detailed explanation, and ChatGPT did better than I expected. It explains the concepts like a human teacher and, most importantly, provides simplified explanations. Not only for academics, but right now, it helps me in learning web development and Python. The ability to explain things makes it smarter than others. Because of ChatGPT, I do not often feel the absence of a human teacher, because it explains the concepts like a teacher. If we use ChatGPT in a better way and efficiently, we can learn everything, including academic and extracurricular subjects. It totally depends on us whether we use it efficiently or not. For me, it is nothing less than a blessing.

2

As a visually impaired student, AI (mainly ChatGPT) has helped me a lot with my academic learning. It gives me solutions for many problems. For instance, when I was preparing for the JEE examination, due to a lack of accessible resources, I found it very difficult to study the tough concepts of PCM. But ChatGPT made it a bit more manageable to understand the concepts. I used to prompt the problem and ask for a detailed explanation, and ChatGPT did better than I expected. It explains the concepts like a human teacher and, most importantly, provides simplified explanations. Not only for academics, but right now, it helps me in learning web development and Python. The ability to explain things makes it smarter than others. Because of ChatGPT, I do not often feel the absence of a human teacher, because it explains the concepts like a teacher. If we use ChatGPT in a better way and efficiently, we can learn everything, including academic and extracurricular subjects. It totally depends on us whether we use it efficiently or not. For me, it is nothing less than a blessing.

1

1 Comment

Hello everyone...... 🚀 Exciting News Alert 🚀 I am working as Junior Data Scientist Intern at Coapps.ai In this internship, I learned modules like Numpy ,pandas ,matplotlib and Machine learning algorithms which are used to expand my knowledge in Data Science domain. As a part of internship, I worked on project called "TRAVEL RECOMMENDATION SYSTEM USING ML". The main aim of the project is to find the user preferences and destination data to suggest personalized travel recommendations. Here is the project code and web application designed by me. Check out the project: https://lnkd.in/gzexYCRn #datascience #technicalskill #recommendationsystem #travelrecommendationsystemusingml #machinelearning #coapps.ai

13

Hello 👋 --- I am very exciteded and I can't wait ... because Tomorrow is my class of iit mandi x MASAI about data anylist experienced real data I am pleasure to learn with high profile professor like miss Indu Joshi and sir Aman Singhal I am only to say thanks to all masai team for launching this type of great course 😊😊#iitmandi #masaischool

4

🏏 #Excited to announce my #latest_project: #Predicting_IPL_Winning_Teams using #AI_ML_models! 📊 Using #diverse_datasets, I've built a #model to #forecast the #outcomes of the #Indian_Premier_League. A huge #ThankYou to InternPe for this #fantastic_opportunity! 🙏 I'd love to hear your #feedback and #suggestions – please share your #thoughts in the #comments below! 💬 #IPL2024 #MachineLearning #ArtificialIntelligence #DataScience #SportsAnalytics #Cricket #TechInnovation #InternPe #AIinSports #PredictiveModeling #BigData #CricketLovers #MLProjects InternPe

18

1 Comment

Interview Experience for AI/ML Engineer Position: Recently, I appeared for an AI/ML Engineer interview, and here's how it went: 1. Introduction & Project Discussion (20 min): The interview started with a deep dive into my professional journey and the key AI/ML projects I’ve worked on. 2. Technical Questions: -- Confusion Matrix: Explained its use in classification tasks and why it's not used for regression. -- Regularization Techniques: Discussed the role of L1 and L2 regularization in preventing overfitting. -- Bias-Variance Tradeoff: Provided an overview of how to balance bias and variance to optimize model performance. -- Transformer Architecture: Walked through the working and components of the transformer model. -- Recurrent Neural Networks (RNN): Explained how RNNs work and where they are best applied. -- Retrieval-Augmented Generation (RAG): Described RAG architecture and its application in enhancing the accuracy of LLMs. Coding Round (20 min): Solved a problem with a focus on AI/ML concepts within the allotted time. Behavioral & Reasoning Round: This round focused on situational judgment and behavioral responses. I believe this is where I could have improved. Result: Not selected. ❌ #InterviewExperience #AIML #MachineLearning #DataScience #CareerGrowth #Learning

5

1 Comment