Portfolio#

Education#

University of Maryland - 09/22 to Present, Expected 05/25

Degree: Bachelor of Science

Major: Computer Science - Machine Learning Track

Minor: Statistics

Relevant Coursework: Artificial Intelligence, Machine Learning, Computer Vision, Natural Language Processing, Data Science, Parallel Computing, Calculus 1, 2 & 3, Statistics, Linear Algebra, Compilers, Computer Systems, Algorithms, Organization of Programming Languages, Object-Oriented Programming 1 & 2, Discrete Math

GPA: 4.0

Experience & Projects#

UMIACS - 10/24 to Present

Undergraduate Research Assistant

Creating a neural video codec to surpass state of the art compression algorithms for image and video data

Models are fit to decode the original video from input pixel coordinates efficiently

Using methods such as model quantization and meta learning to achieve ideal reconstruction quality with high compression, high encoding, and high decoding speeds

https://www.umiacs.umd.edu/
Shahoveisi Lab - 02/24 to 11/24

Undergraduate Research Assistant

Created manuscripts for machine learning research projects related to identifying and managing turfgrass related diseases

Used methods such as transfer learning and gradual unfreezing to train highly accurate nematode image classifiers

Performed automatic hyperparameter optimization using Ray Tune to train scikit-learn and PyTorch models to achieve highest metrics

Performed parallelized automatic image dataset preprocessing using OpenCV and NumPy

https://sites.google.com/view/umdturfgrasspathology/home
SimpleTensor - 02/24 to 05/24

Created a library which provides Tensors with reverse-mode automatic differentiation capabilities using only numpy arrays for the Intro to Artificial Intelligence (CMSC421) class

Supports many differentiable n-dimensional tensor operations such as matrix multiplication, convolution, element-wise functions, aggregate functions, and arithmetic operations, with support for operations along any axes

Created MNIST demo using convolutional, dense, and normalization layers and used techniques such as Xavier/Glorot initialization and residual connections

Fully documented using sphinx at https://vikramrangarajan.github.io/SimpleTensor/

https://vikramrangarajan.github.io/SimpleTensor/
A.M. Best Rating Services - 06/23 to 01/24

Data Strategy Engineer

Gained advanced experience with relational databases, Docker, Linux, Python, and Pandas

Learned to use Azure Data Factory (ADF) to transform and move data on the Azure Cloud Platform

Used Apache Airflow to orchestrate ETL pipelines between on-prem databases and Azure

Accelerated a data pipeline’s execution time from 90 minutes down to 6 minutes using ADF

Technical Skills#

Programming Languages

Technologies

Awards & Certifications#

Astronomer Certification for Apache Airflow Fundamentals - 02/24
https://www.credly.com/badges/82aab031-8123-40de-b310-0c73394b5329/public_url
UMD Computer Science Semester Academic Honors - Fall 22 - Spring 24