Dharun Ramesh

Smart student studying in academic discipline and looking for related experience. Willing to listen and learn and excited to contribute in real-world environment to help find practical, actionable solutions for current needs.

• AI for India 1.0 - GUVI Geek Networks, IITM Research Park
• 100 Days of Code - The Complete Python Pro Bootcamp 2022 - Udemy
• Data Structures and Algorithms - Udemy
• Certificate for Completion of Advanced Cpp Training - Spoken Tutorial
• Certificate for Completion of Python 3.4.3 Training - Spoken Tutorial
• Certificate for Completion of C Training - Spoken Tutorial
• Python - HackerRank
• Certificate of completion - Mathematics - 365 Data Science
• The Complete Machine Learning Course with Python - Udemy
• SQL - HackerRank
• AI for CyberSecurity IBM Qradar - SmartInternz

HERO Campus Challenge Season 7, 2022

• Our team "Bike-O-venters" were grand finalists in the competition.
• Top 5 teams in the Engineering Track among the 53,947 registered for the competition.

HACKHUB 2022

• Led and worked with our team "Team Zero"
• We were amongst the top 15 teams in the competition.
• Made a python application which uses the Pomodoro technique to help users with productivity.

Humanoid Flash VITCC

• Worked with the team "Sherlock Homies" which won 3rd place in the competition.
• We came up with an idea for a robot which performs household tasks.

Dark Patterns Buster Hackathon 2023

• Competition conducted by the Ministry of Consumer Affairs in collaboration with IIT BHU.
• Led our team named "Interface Inquisitors" to the grand finale of the competition.
• We worked on an application which has a web extension which identifies dark patterns on the interface e-commerce websites for deceitful market practices such as forced continuity, etc.

TransFusion-UFNet: A Self-Attentive Transfer Learning Architecture with Uncertainty Fusion

• An auto-encoder which performs audio super resolution.
• Worked on the project during the Samsung PRISM internship

InSync: An application for the neuro-diverse

• Worked and led this project during the NUS education programme.
• Helps identify sounds which trigger people with Audio Processing Disorders and identifies the sounds and deploys methods to help intervene and address the discomforts caused by the triggering sound.
• Developed an facial emotion recognition model and an audio classification model using YAMNet as key components of the project.

Optimized Ensemble Deep Learning for Image Segmentation with Uncertainty Quantification

• Use of pre-trained models like ResNet50 and EfficientNetB7 for feature extraction.
• Implemented model pruning to reduce complexity and improve inference speed.
• Used polynomial decay for pruning schedule to gradually increase sparsity.