May 2022 – Jul 2022 · 2 mos · Bengaluru, Karnataka, India

• Offered PPO
• Estimated default probabilities of sub-standard agricultural lendings and reduced costs of recovery by analyzing the loan portfolio of the bank
• Created an end-to-end pipeline in SAS by consolidating the 0.48Mn+ loans dataset from Oracle databases to perform multivariate analysis
• Integrated farm Pincode data into a decision tree model that assigns a risk score to each farm and identifies the probable loan defaulters
• Achieved an AUC of 0.7 to perform collection analytics by estimating the probabilities of default using a step-wise logistic regression model
• Reduced the recovery cost by classifying defaulters into 3 classes for suitable collection measures using ensemble boost

May 2021 – Jul 2021 · 2 mos · Kolkata, West Bengal, India · Remote

• Extracted 2.5M time series data from AMFI for 900+ mutual funds over 10 years using RPA and calculated 8 risk-return ratios
• Examined 90+ annual reports of 14 AMCs to calculate and backtest portfolio VaR and CVaR of 120+ schemes using the Kupiec test
• Factor clustered funds using t-SNE to group into risk baskets and conducted a cross-sectional analysis to monitor volatility changes
• Brokered data and insightful analysis to an investment research firm ”Value Research”, leading to a revenue increase of INR 0.2M

Nov 2020 – Feb 2021 · 3 mos · Taipei City, Taiwan · Remote

• Applied Clustering Analysis Methods on CPS PISA 2015 Data to map the patterns in Data for cultural contexts
• Analyzed Collaborative Problem Solving abilities for 5904 regions based on 4 topics extracted from 120 questions using Topic Modelling
• Utilised ANOVA to check for significant statistical differences in the scores of 56 countries based on their development and geography
• Achieved silhouette coefficient of 0.63 by studying the dendrogram and developing a hierarchical clustering model of optimum clusters
• Trained and visualized a decision tree model on clustering results to interpret the clustering process using dtreeviz and studied outliers
• First authored a research paper based on the findings, currently under review process in the British Journal of Educational Technology

Apr 2020 – Oct 2021 · 1 yr 6 mos

• Spearheaded a team of 10 members to develop the official registration website for participants of Spring Fest, visited by over 9000+ users
• Responsible solely for developing and managing the Merchandise website, which subsequently increased merchandise sale by 21% YOY
• Carried out 50+ code reviews, identified and corrected 10 bottlenecks; 100+ bugs to ensure code quality for the production deployment
• Benefitted and trained 400+ campus students in web development skills by conducting spring fest summer web development workshop

Jul 2019 – Apr 2020 · 9 mos

• Being a Web Sub head, I successfully developed a website for Nation Wide Prelims of Spring Fest 2020. It is being developed in React js with a php and SQL supported framework.It has amazing events and secured database.

Feb 2020 – Jul 2020 · 5 mos · Beijing, China · Remote

• 1. Conceptualized and Built from Scratch: Developed a robust Hyper-Video Platform for Online Learning using PHP and MySQL, designed to integrate seamlessly with the Moodle platform for future scalability.
• 2. Cost-Effective Deployment: Successfully hosted the platform on low-cost servers, optimizing resources without compromising functionality or user experience.
• Real-Time Student Monitoring System: Engineered a tracking mechanism to monitor student activities in real time, including video watch duration and engagement, with all data securely stored in a dynamic database.
• 3. Professor Dashboard Development: Designed an intuitive dashboard for professors, offering insights into student activity, engagement metrics, and course-specific performance, enabling data-driven decisions.
• Interactive Learning Features:
• 1. Empowered professors to create courses, upload videos, and track student participation.
• 2. Incorporated a real-time chat system where students could comment, react, and add personalized notes at specific video timestamps.
• 3. Embedded quizzes directly into video lectures, introducing gamified learning with student rankings based on scores and interactivity.

May 2018 – Jan 2020 · 1 yr 8 mos · Indian Institute of Technology, Kharagpur

• Proposed a Non-Invasive Health Monitoring System: Using photo spectroscopy to analyze blood components and photoacoustics for glucose concentration detection, alongside scattering spectroscopy to measure other critical parameters such as haemoglobin levels, oxygen saturation, and cholesterol, without needing blood extraction. Enabling early detection and prediction of diseases to promote healthier lifestyles.