Girievasan Manivannan

Experience

• 

  National Institute of Technology, Tiruchirappalli

  Jan 2021 - Oct 2021

  Research Internship

  • Accomplished a notable 15% reduction in Mean Squared Error and outperformed previous methodologies for channel estimation by implementing radial basis interpolation on a sparse input dataset for the ChannelNet deep learning architecture.• Optimized the ChannelNet deep learning architecture for channel estimation and displayed superior results through the development of a 20-layer Very Deep Super Resolution (VDSR) architecture leading to improved interpolation of sparse channels.

• 

  University College Dublin

  Feb 2022 - Dec 2022

  Research Internship

  • Enhanced diabetes diagnosis and recorded an outstanding accuracy of 95.5% and a recall of 100% by utilizing the Support Vector Machines and Random Forests Classifiers exhibiting adept knowledge in Machine Learning toolkits and algorithms.• Optimized machine learning algorithm efficiency by 15% and mitigated data imbalance in medical dataset through strategic implementation of SMOTE technique and bolstered overall algorithm performance by a significant 10%.

• 

  National Institute of Technology, Tiruchirappalli

  Jan 2023 - May 2023

  Final Year Research Project

  • Took part in a six-month project on optimizing electric vehicle charging station cost by providing strategic direction, resolving challenges, ensuring team’s productivity and adherence to project goal demonstrating resilient leadership and teamwork skills.• Leveraged advanced machine learning toolkits and techniques to devise a short-term forecasting model to predict load demand at EV stations and enhanced the efficiency of the predictive model by 8%. •Architected an innovative scheduling algorithm, precisely allocating charging intervals for electric vehicles, leading to a significant 10% reduction in charging costs. • Published this research work at the IEEE International Conference on Energy Technologies for Future Grids displaying strong research communication skills towards a public audience of peers.data. Show less

• 

  Rutgers University–New Brunswick

  Sept 2023 - Dec 2023

  Voluntary AI Research

  • Managed a 3-member team by defining roles and responsibilities, providing feedbacks, planning and executing goals to ensure a cohesive workflow towards the development of a lightweight Deep Learning architecture.• Architected a 10-layer light weight neural network for defect classification in solar cells on hardware-constrained devices through knowledge distillation and leveraged GAN networks for data augmentation to enhance efficiency of neural networks.• Optimized deep learning model performance by 7% upon fine-tuning on model’s parameters, achieving an outstanding F-1 score of 88, Recall of 87, and an accuracy of 85, showcasing proficient expertise in deep learning model development. Show less