SGSumit Gupta

Sumit Gupta

*This is an algorithmically generated author record based on publicly available information.

Sumit Gupta has a total of 11 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • AI-Driven Mock Interviews: A Catalyst for Enhanced Interview Performance

    by Abinav Dayal, Jayasri Angara, Ranbir Sinha, Sinha Research, Sumit Gupta, Ravi Shankar Saripalle, Sreenu Ponnada

    Mock interviews have long been recognized as a valuable tool for job seekers to enhance their readiness for real job interviews, providing invaluable insights into a candidate's communication, confidence, preparation, and time management skills. However, traditional human-tohuman mock interviews can be expensive, offer limited attempts, may not always provide immediate feedback, and could potentially be influenced by personal biases. Additionally, they may not always provide the diverse range of interview scenarios that a candidate might encounter in the real world. In order to study the impact of AI-Powered mock interviews, this research utilizes Skill2030.com a platform that offers job seekers a realistic AI-powered mock interview experience with personalized feedback while promoting diversity and inclusivity. An ethical approach is followed by the researchers in utilizing AI by excluding protected parameters like age, gender, ethnicity, etc., from the analysis. The study involves 6619 interviews conducted for candidates from 998 colleges across India representing all tiers (Deemed Universities, IITs, Affiliated, and Autonomous Engineering Colleges) with a 30-70 rural/urban split. With a high positive feedback rate of 98.5%, the AIbased platform successfully provides realistic simulations, personalized feedback, and greater accessibility, enhancing overall job seekers' preparedness and success rates in India's competitive job market. Results gleaned from the feedback demonstrate that candidates with robust communication and writing skills exhibited a higher degree of confidence. A separate study of 500 candidates who used the AI-based platform as a learning tool to enhance their interview performance reveals promising and interesting patterns in how candidates improve by over 30% with consistent practice in parameters like technical correctness, pronunciation, intonation, fluency, and grammar.


  • Compression Strategies for Massive Natural Language Generation Models

    by Abhinav Dayal, Jayasri Angara, Sumit Gupta, Ravi Shankar Saripalle

  • QuMaDe: Quick Foreground Mask and Monocular Depth Data Generation

    by Sridevi Bonthu, Abhinav Dayal, Arla Lakshmana Rao, Sumit Gupta


  • Machine learning-based authorship attribution using token n-grams and other time tested features

    by Sumit Gupta, Swarupa Das, Jyotish Ranjan Mallik

    Authorship Attribution is a process to determine and/or identify the author of a given text document. The relevance of this research area comes to the fore when two or more writers claim to be the prospective authors of an unidentified or anonymous text document or are unwilling to accept any authorship. This research work aims to utilize various Machine Learning techniques in order to solve the problem of author identification. In the proposed approach, a number of textual features such as Token n-grams, Stylometric features, bag-of-words and TF-IDF have been extracted. Experimentation has been performed on three datasets viz. Spooky Author Identification dataset, Reuter_50_50 dataset and Manual dataset with 3 different train-test split ratios viz. 80-20, 70-30 and 66.67-33.33. Models have been built and tested with supervised learning algorithms such as Naive Bayes, Support Vector Machine, K-Nearest Neighbor, Decision Tree and Random Forest. The proposed system yields promising results. For the Spooky dataset, the best accuracy score obtained is 84.14% with bag-of-words using Naïve Bayes classifier. The best accuracy score of 86.2% is computed for the Reuter_50_50 dataset with 2100 most frequent words when the classifier used is Support Vector Machine. For the Manual dataset, the best score of 96.67% is obtained using the Naïve Bayes Classification Model with both 5-fold and 10-fold cross validation when both syntactic features and 600 most frequent unigrams are used in combination.