HRIDITA RUBYAT
**** ****** **** **, ********* - *8262, NC
(717) -863-0511 ad5okf@r.postjobfree.com Porfolio/BArch Certifications Linkedin Github link Tableau Public
EDUCATION:
Master of Science Data Science and Business Analytics 08/2023 - 05/2025 (Expected) University of North Carolina at Charlotte, Charlotte, NC GPA: 4.00/4.00 Bachelor of Architecture 2020
Bangladesh University of Engineering & Technology, Dhaka, Bangladesh TECHNICAL SKILLS AND TOOLS:
Python, Pytorch, Streamlit, OpenCV, Yolo, R, SQL, PostgreSQL, Tableau, SAS 9.4, Git/Github, Matplotlib, Power BI, Seaborn, Microsoft Word, Microsoft Excel, Microsoft Powerpoint. PROJECTS:
Applied ML: Detection of Hidden Data Embedded with Digital Media 01/2024-05/2024
Python, Pytorch, Streamlit, PIL, Scikit-learn, EDA, Transfer Learning
● Employed Advanced Machine Learning and Deep Learning Techniques.
● Implemented multiple CNN models (Handbuilt CNN, Resnet-18, Efficientnet -B0, B2, B3) to Detect Hidden Data through Image Steganography.
● Used Pre-Trained Network weights, Binary/Multiclass Classification, Data Manipulation and Data Augmentation Techniques initially.
● Key outcome: Empowered Forensic investigators in recognizing covert communication activities. Achieved 80% Accuracy by retraining Effnet- B0, taking a non-conventional approach. Built a Streamlit App where User can watch Live Demo of Hidden Data Detection. Advanced Business Analytics: Stock Analysis 01/2024-05/2024
Python, Statsmodels, Scikit-learn, Keras, yfinance
● Performed Exploratory Data Analysis, Data Cleaning, and checked Data Quality.
● Non-Time series Model (Random Forest with Feature Engineering) and Time-Series Models
(Linear, Polynomial Models, LSTM, SARIMA) used as Analytical Techniques to Forecast short term Future Stock Price for Microsoft.
● Explored Trend, Seasonality, Volatility, Statistical Significance and tuned Hyperparameters to get the best metrics.
● Key outcome: Compared Model performances and Recommended best Stock. LSTM Model achieved least error
(4.26%).
Deep Learning/ Applied Computer Vision: Skin Cancer Classification Python, Pytorch 01/2024-05/2024
● Used Hand-Built VGG Neural Network architecture and Data Augmentation Techniques.
● Implemented a Pretrained Resnet-18 Model for Transfer Learning as a Feature Extractor.
● Compared Network performances and generated Gradient Class Activation Maps.
● Key outcome: Pre-trained model outperformed (82%) the hand-built CNN (80.2%) in terms of Accuracy, Precision, Recall and F1-score.
Applied Computer Vision: NFL Defender Identification Python, cv2 01/2024-05/2024
● Utilized Object detection, and Object Tracking for automating the identification. when an NFL defender is able to get past a blocker after the ball is snapped.
● Key outcome: Reduced the manual effort of coaches previously required during tape analysis. World Happiness Database Design MySQL Workbench, Draw.IO 8/2023-12/2023
● Developed and designed a normalized database for World Happiness Report.
● Led the development of the "Blissify App" to address diverse user segments.
● Created a Use Case for expansion of the existing dataset to support a new feature, Entity Relationship Diagram.
● Performed query indexing and optimization using SQL, created complex queries, stored procedures, triggers, and views.
● Key outcome: Provided evidence-based happiness analytics for informed decision-making to users worldwide. Movie Dataset Visualization and Storytelling Tableau, Python 8/2023 -12/2023
● Marvel vs DC Superheroes Bankability
● Analyzed the Box-office Movie Dataset to identify insightful information and presented through storytelling.
● Designed and developed interactive Tableau dashboards to visualize interesting characteristics within the data. Credit Card Approval Prediction R, earth, knn, nnet 8/2023-12/2023
● Used techniques like Linear Regression (experimented with multiple variables and Polynomial terms), Non-Linear Regression (MARS) and Non-Parametric Regression KNN) to capture the relationship between independent and target Variables.
● Utilized k-fold Cross Validation to train and assess the models.
● Key Outcome: MARS model demonstrated the highest Predictive Accuracy (lowest RMSE value 39% among all). ACADEMIC COURSES:
● Database Systems for Data Scientists ● Applied Computer Vision ● Cloud Computing Fall 2024
● Business Intelligence and Analytics ● Applied Machine Learning ● AI and Deep Learning Fall 2024
● Strategic Business Analytics ● Advanced Business Analytics ● Supply Chain Management Fall 2024
● Visual Storytelling and Analytics
JOB EXPERIENCE:
Graphic Designer Color Wizard, Bangladesh 1/2020 – 12/2021
● Conducted creative tasks including Logo Design, Social Media Ad Development. Conceptualized designs aligned with client’s brand identities and marketing objectives.
● Demonstrated Strong communication, Critical thinking, Business acumen and quick learning ability.
VOLUNTEERING EXPERIENCE:
● Speaker Host and Volunteer at Women in Data Science Conference, Charlotte, NC. 03/2024
● Speaker Host and Volunteer at Analytics Frontier Conference, Charlotte, NC. 04/2024