Hi, this is Kush

University of Massachusetts, Amherst

Pursuing my BS - CompSci Degree from the Manning College of Information and Computer Sciences at UMass, Amherst. Minor in Psychology.

Associated with the College of Natural Sciences for the Research-Based Integrated Concentration in Science curriculum, where I have already published 4 projects on the official UMass website. The curriculum is backed by various organizations, including the National Science Foundation.

Lastly, I am also enrolled in the Commonwealth Honors College.

Reference Links

• https://orcid.org/0000-0003-0656-5478
• Compiled List of Published Works under iCons: https://suobset.github.io/iCons
• Relevant Coursework"
  • Data Structures
  • Computer Systems' Principles
  • Reasoning Under Uncertainty
  • Programming Methodology
  • Algorithm Design
  • Artificial Intelligence
  • Operating Systems
  • Information Retrieval & Search Engines
  • Compilers and Networks

Integrated Concentration in STEM @ College of Natural Sciences, UMass Amherst

• Integrated Concentration in Sciences; a 20-Credit Research Based Undertaking backed by NSF.
• Currently working on Excess HVAC Energy Usage due to Inaccurate Thermostats in Commercial Buildings.
• Past Projects include Accessibility Analysis of Boston T (with Museum of Science), along with a couple of Independent Research Projects linked above.

CICS Systems Lab

• Currently working on RUST Compilation Performance Metrics to compare and contrast the performance with other Programming Languages.
• Creating Tracers on Python, RUST, and C/C++ to check how many execution steps are needed per step, along with real-time ongoing complexity metrics.
• Primarily comparing the generated x86 Assembly from different Compilers, and will be using Gaussian Mixture Models in the future to compare different languages across a number of runtimes for different programs.

LinKaGe Lab: Smith College

• Assisting in the Software Development part of the current research under the guidance of Dr. Ileana Streinu at LinKaGe Lab.
• Optimizing systems and servers to ensure the lab's software programs work efficiently, as well as ensuring upkeep, and security.
• Technologies in use: C, Java Servlets, and Python, along with a heavy utilization of UNIX CLI programs and server systems.

FN MathLogic Consulting Services

• Constructed, trained, analyzed, and optimized Neural Networks on tabular, and image based datasets via Tensorflow and Keras.
• Achieved a training accuracy of 96.21% and testing accuracy of 95.17% on the MNIST Database. Similar figures were achieved on the VisualQA database, as well as multiple tabular datasets reflecting Market Conditions provided by the company.
• Experience with different Neural Networks, Gradient Descent, Loss Curves, Precision vs. Recall, and coding the same in Python 3, along with Tensorflow, Keras, Pandas, Numpy, and Sci-Kit.
• Performed Data Roll Up at Customer Level, Resource Allocation, Neural Network Construction and Optimization, and Ensemble Machine Learning for predicting Credit Card Defaults in a Kaggle Competition with American Express with a 92% accuracy.

Spider Business Consultancy

• Ideation, design, and preliminary implementation of a Business Assistant's model to cover over 19 countries in 5 continents.
• On The Move: Product Page.

The Indian High School

• Led a team of 14 graduates to develop a Mobile App integrated with Microsoft 365 to move classwork and exams for 4500+ students online. Utilized Kotlin, Swift, Firebase, and Flutter. Trained 150 Faculty Members on using the same in 5 days.
• Implemented the authentication system of the school's newly developed Alumni Portal using Django, Python, and SQL

Tata Consultancy Services, Dubai

• Designed a platform in 6 weeks for a Food Company (Kout Food Group) to digitize the infrastructure across 19 branches.
• Utilized Amazon Web Services, SQL, and Hadoop to aid in the development of a web-based platform which manages the inventory and daily activities. Aided the development of real time tracking services as well.
• Deployed the initial phase of the product in under 2 weeks, as well as devised a 6-month migration plan to turn 100% digital.

Five College ASA Datafest | First Place

• Did a in-depth Analysis of a Dataset consisting of conversations and data between client-company interactions.
• Created an elaborate system for Information Retrieval and Extraction via NLP Concepts (Tokenization, Stopping, Stemming, Vectorization, and Text Extraction).
• Further Details soon.

Mapping Transit Accessibility in Boston

• Collaborated with a team of 5 people to map the Boston Public Metro System, and analyze its accessibility.
• Programmed Interactive Map using GIS and R to showcase accessibility of 162 metro stations, along with real-time on-field data taken in Boston over the span of 2 weeks, backed by MBTA, The Museum of Science at Boston, and UMass CNS.

Assessing the Development of Renewable Energy

• Launched an interactive website to showcase the development and adoption of Flow Batteries between 2015, predicted until 2027.
• Built interactive graphics using JavaScript, and gave a 10-minute keynote to help the audience interact and experiment with the data.

MoodMusic: React app developed at HackUMass IX (48hr Hackathon)

Application which integrates with a user's Spotify account to determine moods of current songs using a Mathematical Model coded in Python. The rest of the Web App uses React.js, and Flask to bridge the gap between programming languages.

DermSafe: Android App Developed at HackUMass VIII

AI-based Android app to determine if an individual has skin cancer based on uploaded pictures (either taken from the camera or uploaded from the phone's gallery). Uses Java, Python, and Tensorflow.