KARTHIKA SELVARAJ
+1-239-***-**** ad3zo8@r.postjobfree.com LinkedIn Github
EDUCATION
Master of Science in data science – CGPA: 3.85 Dec 2023 Seattle University
Relevant Courses: Statistical Machine Learning, Probability for DS, Data Visualization, Intro to Deep Learning, Bigdata, Numerical Methods
Bachelor of Engineering in Electronics & Communication– CGPA: 3.5 Apr 2016 Anna University
SKILLS
Programming: python, R, SAS, C# Machine Learning: Pytorch, Keras, Tensorflow, NumPy, Pandas, Matplotlib MLOps: Git, Spark, DBs: SQL,Hadoop NLP / LLMs: Hugging face, NLTK Cloud : AWS Visualization : Tableau, PowerBI
PROFESSIONAL EXPERIENCE
Seattle University Seattle, WA
Data Analytics Oct 2023 – Dec 2023
● Conducted hypothesis testing to assess the impact of the employer engagement score on campus recruiter performance, resulting in a 20% increase.
● Automated reports with VBA, saving 12 hours/week for strategic engagement.
● Effectively communicated technical insights, leading to a 15% increase in engagement. Seattle University Seattle, WA
Research Assistant Jun 2022 – Jan 2023
● Collaborated with executives to analyze stellar cluster properties using Bayesian Statistics.
● Utilized Gaussian Kernel density for cluster type determination from photometric data.
● Published a detailed research report on the experimental design in the AAS Conference. Camp Korey. Seattle, WA
Data Scientist(Project Coursework) May 2022 – Aug 2022
● Collaborated to optimized donor management statistically for improved fundraising.
● Established an engagement score through complex queries and KPIs from Customer Perspective
● Utilized MySQL and PowerBI for data exploration, providing actionable insights.
● Developed predictive ANN models with an RMSE value of 1.93 for forecasting donor engagement scores. IBM Chennai, India
Senior Data Analyst Aug 2016 – Apr 2021
● Designed and implemented a tool using MySQL and Unix shell script for post-migration data verification, generating end-user reports.
● Streamlined data loading into a star schema using SSIS Mappings and Reusable Transformations for efficient processing.
● Created interactive Power BI reports and dashboards, facilitating data-centric decision-making for clients and internal stakeholders.
● Established a fail-safe batch processing pipeline using Spark RDDs and data frames for data extraction, transformation, and loading.
RESEARCH PROJECTS
Amazon Product Recommendation Using LLMs
• Utilized advanced alignment techniques, including Supervised Fine-Tuning and Reinforcement Learning with Human Feedback, in conjunction with Collaborative Filtering, to tailor and optimize individualized product recommendations on the Amazon platform.
• Demonstrated proficiency in implementing LLMs, specifically the T5 model, to enhance accuracy and relevance in the Product Functionality
Identifying the Bird Species by their Call Using CNN
● Implemented CNN algorithms for bird species prediction, achieving 96.7% accuracy with a simpler binary classification model (convolution layer: 32,64; dense layer: 32,1) in 26 sec.
● The multi-class model with required more layers (convolution: 32,64,128,256; dense: 512,256,12), achieving 71% accuracy in 8 minutes, balancing accuracy, and computing speed. Using Decision Tree to Predict Teenage Substance Abuse Factors
● Utilized feature selection and ensemble methods, including bagging, and boosting, to enhance model performance.
● Achieved a 91% accuracy rate for binary classification of analysis. Multi-class classification using random forest predicted consumption risk with 47% accuracy. Regression model with boosting accurately predicted marijuana usage frequency