Rajkumar
Conjeevaram
Mohan
Data Scientist
As a graduate student in Data Science, I possess extensive knowledge of the sub- ject and have hands-on experience in implementing Statistical Machine Learning and Deep Learning models. Throughout my academic journey, I have undertaken various challenging projects covering domains such as finance, healthcare, and more. I leverage my expertise in programming languages like Python and R, along with my proficiency in SQL, to design and implement predictive models, conduct exploratory data analysis, and develop data-driven solutions for complex prob- lems. Apart from Machine Learning, I am also passionate about other branches of AI and constantly strive to learn more. I approach challenges with enthusiasm and take pride in my ability to adapt quickly to new technologies and environments. Rajkumar, who is passionate about the subject, and is self-driven, would prove to be an exceptional candidate for the Data Scientist position. Personal Info
Phone
adzgzx@r.postjobfree.com
Date of birth
01-18-1990
Skills
R
Python
Deep Learning
PyTorch
TensorFlow
Numpy
Scikit-Learn
Tableau
AWS
GCP
Analytics
MySQL
MongoDB
Neo4j
Linux
Algorithms
Experience
Student Technical Support Specialist
George Washington University / 06/2022 - 12/2022
Served as a Data Analyst, whose primary role involved analyzing data that had complex relationships, cleansing it, and creating interactive visualizations that effectively conveyed important information about the inter-group relation- ships among militant organizations worldwide (e.g. how they evolved over time). Through my work, I improved the system's efficiency by approximately 30%, as measured by the latency in loading different visualizations. Complex visualization of the network was one of the primary objectives of the work. This was a United States Department of Defense project funded by the DHS. Education
MS Data Science
George Washington University / 12/2021 - Present
Washington, DC
Notable courses: Machine Learning, Time Series Analysis & Modeling, Deep Learning, and Natural Language Processing.
MS Artificial Intelligence
Imperial College London / 07/2015 - 10/2017
London
Notable courses: Advanced Statistical Machine Learning, Computer Vision, and Intelligent Data Analysis & Probabilistic Inferences. BSc (Hons) Computer Information Systems
University of Liverpool / 07/2010 - 06/2013
Liverpool
First Class Honors Degree
Certifications
TensorFlow Developer
Coursera / 10/2020 - Present
Natural Language Processing
Coursera / 12/2020 - Present
Practical Time Series Analysis
Coursera / 11/2021 - Present
Machine Learning
Statistics
NoSQL
Docker
Git
SciPy
NumPy
Jupyter
Pandas
Matplotlib
Seaborn
Excel
Keras
PySpark
Links
https://www.linkedin.com/in/rajkumarcm
GitHub
https://www.github.com/rajkumarcm
Technical Projects
Brain Tumor Segmentation
03/2023 - 04/2023
Washington, DC
Classified pixel-wise tumorous cells in the brain, and obtained 0.78 Jaccard score using 3D UResNet.
Challenges and Solutions:
1. High-dimensional images caused memory problems (solution: Grouped con- volution)
2. Low performance (solution: Residual blocks)
Brain Tumor Recognition
10/2022 - 11/2022
Washington, DC
Recognized presence of tumor in the brain using large scale Resnet, and achieved 94% F1-macro on test set.
US Air Pollution Prediction and Forecast
09/2022 - 11/2022
Washington, DC
Estimated the order of the AR, and MA processes to predict and forecast the Carbon Monoxide (CO) Air quality index, and achieved 68.11% R2 value using ARIMA, and Holt-Winters Seasonal Method.
Challenges and Solutions:
1. Missing Data - (Seasonal Naive method was used to replace the missing values) 2. No adequate information to predict CO AQI - (Used weather information to assist in predicting CO AQI)
3. Seasonal data - (Log transformation followed by 1st order differencing to make it stationary)
4. Mix of AR and MA processes - (Having only one of these makes it easier to trace the order of the process through its respective plot. However, when there is a mix of processes that generated the data, I had to resort to the GPAC method that help determine the mix order)
Credit Card Default Prediction
03/2022 - 04/2022
Washington, DC
Classified Taiwanese bank clients who have the potential to default, with 0.813 F1-score using Decision Tree.