Jun 2021 – Jul 2021 · 1 mo

• Trade Promotion Optimisation
• Developed a promotion model with potential 10% increment in wholesale revenue impacting INR 50 million. Identified maximum impact clusters via Exploratory Data Analysis; presented findings to P&G India IT leaders. Performed feature engineering, retaining impactful features to prevent overfitting and maintain model robustness. Also collaborated with business team to obtain constraints on promotions to ensure it can be smoothly deployed

Apr 2021 – Mar 2023 · 1 yr 11 mos

• An interdisciplinary team of 50+ students that works on designing space systems for CubeSats. With this objective Antenna Deployment System and Star Tracker based Attitude Determination System are being developed. They are aimed to be space-qualified and used for further missions by the team. The team is also part of the Great Lunar Expedition for Everyone which is a global initiative led by the Space Grants of the University of Colorado. The team also works on creating a software framework for Rendezvous and Docking simulations for CubeSats. Ham radio club of the institute also functions under IITBSSP where we conduct amateur radio club activities.

Sep 2020 – Mar 2021 · 6 mos

• Heading a team of 50 students, working on interdisciplinary space systems for CubeSats.

Feb 2019 – Sep 2020 · 1 yr 7 mos

Jun 2020 – Jan 2021 · 7 mos · Delhi, India

• License Plate Recognition
• Designed an end-to-end pipeline for License Plate Recognition in Indian context from CCTV images. Developed a 90% accurate license plate recognition system on-par with commercially available ALPR systems, which works on tollbooth images using a Single-Shot Detector followed by a custom sequence recognition CNN. Improved plate recognition model performance by 15% via a novel method for pre-training using synthetic plates. Achieved 5% overall increment in model performance via hyperparameter tuning and image augmentation methods.
• Weapon Recognition using CCTV Images
• Worked on developing a system for weapon recognition using CCTV footage by fine-tuning EfficientDet model. Created a synthetic weapon detection dataset using Pillow and albumentations for detection training.

May 2020 – May 2020 · 0 mo · Bangalore

• Worked as part of the Flight Control Simulations team to automate Simulink model generation contributing to cost reduction.

Apr 2019 – Jan 2020 · 9 mos · Mumbai

• Recruited seven international ambassadors for TechFest World MUN who are responsible for
• improving the MUN's outreach in their country and recruiting quality delegates for the conference. Compiled databases of existing MUNs and contacted them to form partnerships and delegate exchanges. Headed a team of 10 organizers for publicity of the conference and coordinating the overall event.