Vamsi Deekshit Kanakavety

Vamsi Deekshit Kanakavety

Computer Science Graduate Student at Arizona State University

Posts by Collection

portfolio

publications

Full Stack Developer

Oracle, July 06, 2021 - June 27, 2023

  • Contributed in the agile software development of a comprehensive end-to-end banking application with over 200 customer banks with more than 1 Million active users, catering to a diverse range of corporate and retail banking needs.
  • Played a key role in the modularization of a monolithic banking application to a microservices-based architecture, enhancing the product scalability, including a 40% boost in system performance, a 30% reduction in update-related downtime.
  • Highly involved in building more than 20 RESTful Web Services using JAX-RS API and Jersey Implementation and integrated multiple User-Interface Components using Oracle JET (JavaScript Extension Toolkit).
  • Significantly contributed to the design and development of multiple User-Interface Components using Oracle JET (JavaScript Extension Toolkit), which greatly improved user experience and application responsiveness, enhancing overall system performance.
  • Engaged in over 20 client consultation sessions, demonstrating a strong commitment to addressing client needs and providing valuable technical assistance and adeptly resolved more than 50 application related bugs.

talks

Blind Navigator

Published:

  • This innovative glove project aids the visually impaired by utilizing specialized sensors to measure distances from obstacles.
  • The glove provides real-time feedback through vibrations, allowing users to navigate urban environments with increased safety and confidence.
  • Its purpose is to enhance the mobility and independence of individuals with visual impairments.

Seizure Detection Using ML Algorithms

Published:

  • A novel approach for the classification of electroencephalogram (EEG) signals for seizure detection is introduced, leveraging both feature selection and channel selection techniques.
  • Channel selection was conducted via the variance method. Features were then extracted from intrinsic mode functions obtained post empirical mode decomposition of the EEG signal. Feature selection involved a one-way analysis of variance test with a preset threshold probability value.
  • The classification of seizure and non-seizure signals is done on selected features using Decision Tree and k-nearest neighbor algorithms.
  • The proposed method achieves an impressive accuracy of 95.6\% in classifying EEG signals for seizure detection, outperforming existing methods.

Branch Location Maintenance

Published:

  • Integrates a Spring-based REST service, containerized using Docker, with an Oracle JET user interface. The primary goal is to facilitate the storage of branch details entered by front-end users.
  • Additionally, the system enables the retrieval of specific branch information, including the visualization of branch locations on a map, leveraging the Google Maps Javascript API.
  • The seamless interaction between the user interface and the backend service enhances the functionality for managing and visualizing branch data efficiently.

UC-Berkeley: The PacMan Project

Published:

  • Implemented various search algorithms, including depth-first, breadth-first, uniform cost, and A*, in the Pacman environment to address navigation and traveling salesman problems. The goal is to obtain optimal paths while strategically avoiding ghosts.
  • Encompassed the duality of Pacman as both an adversarial and stochastic search problem. Implemented multiagent minimax and expectimax algorithms, as well as the creation of evaluation functions to improve Pacman’s decision-making abilities in the game.
  • Implemented value iteration and Q-learning reinforcement learning algorithms, applying them to the AIMA textbook’s Gridworld, Pacman, and a simulated crawling robot. Also developed an approximate Q-learning agent using feature extractors, achieving successful learning outcomes in various environments.
  • Developed classification algorithms using TensorFlow, including gradient descent and stochastic variants for neural network training. Optimized model performance through thoughtful feature design, emphasizing the importance of selecting and extracting relevant features for improved classification accuracy.

Credit Card Default Prediction

Published:

  • Developed a comprehensive credit card default prediction system employing machine learning algorithms such as Logistic Regression, Random Forests, Support Vector Machines (SVM), and Deep Neural Networks (DNN).
  • Conducted thorough data exploration and preprocessing to identify crucial features and patterns related to credit card defaults.
  • The model strives to be robust and interpretable, contributing to consumer protection, enhancing credit risk management practices, and fostering trust within the financial industry.
  • Emphasized the importance of model robustness and interpretability in addressing multifaceted challenges inherent in credit default prediction.

teaching

Master of Science, Computer Science

Arizona State University, August 2023 - Expected May 2025 , GPA 4/4

Coursework : Artificial Intelligence, Statistical Machine Learning, Foundations of Algorithms, Cloud Computing, Knowledge Representation and Reasoning, Data Visualization, Data Structures and Algorithms.