I have 5+ years of work experience as a Data Scientist. With a Master of Science in Data Science from Northeastern University and a robust background in Machine Learning, Deep Learning, and predictive analytics, I am enthusiastic about the opportunity to contribute my expertise to your team.

During my tenure as a Data Scientist Intern at Spare-it, I spearheaded instance segmentation projects using cutting-edge computer vision techniques such as YOLO, U-NET, and Mask R-CNN for sustainable waste management. This experience honed my skills in object detection and extraction, AWS CDK implementation, and real-time data visualization for actionable insights.

My tenure at Datametica further amplified my capabilities. Notably, I led initiatives for BestBuy that resulted in a 7% increase in Average Order Value and a 12% improvement in Return on Ad Spend through personalized product recommendations and behavioral analytics. Additionally, my work for Discover Bank showcased my proficiency in text classification and summarization using attention-based transformer models, significantly improving ticket classification accuracy and legal document summarization.

At every stage of my career, I've engaged with diverse stakeholders, collaborated closely with cross-functional teams, and presented data-driven insights to drive key business decisions. My technical acumen spans a wide array of tools and platforms, including Python, SQL, Numpy, Pandas, Tableau, AWS, PySpark, and more.

Machine Learning, Predictive Analytics, Deep Learning, Natural Language Processing, Computer Vision, Hypothesis Testing.

Northeastern University

Instance Segmentation using Computer Vision for sustainable waste management

*   Object Detection and Segmentation using YOLO, U-NET, and Mask R-CNN: Performed Instance segmentation by fine-tuning pre-trained models like Circular-Net on waste management dataset. Developed Lambda functions using AWS CDK to facilitate data pipelines.
*   Extracting object-info to generate reports: Extracted image objects labels and its coordinates from output COCO files obtained from Circularnet in order to generate daily dashboard reports highlighting the contamination report and diversion rate in the waste disposal.

Data Scientist Intern

1.  \[BestBuy\] - Personalized Product Recommendation for Revenue Optimization

  

*   Content based Product Recommendation: Developed content-based recommendations based on product features and profile personas. Also, developed collaborative models using user-product interactions leveraging Matrix Factorization and Singular Value Decomposition.
*   KPIs and Stakeholder Interaction: Interacted with Product and Marketing teams for communicating insights and monitored KPIs by improving metrics like Average Order Value by 7%, Click Through Rate (CTR) by 9%, Return on Ad Spend by 12%.

  

*   \[BestBuy\] - Behavioral Analytics to enhance Customer Satisfaction, Brand Loyalty, and Churn Reduction
*   Social Media Engagement: Analyzed user engagement with Facebook, Twitter, and other platforms by tracking likes, shares, comments to understand customer sentiment, satisfaction, to help improve design and response times.
*   Clickstream Analytics for Churn Reduction: Implemented granular analysis by mapping customer journey, path-to-purchase trails to identify key touchpoints, page layouts and Call-to-Action (CTA) placements influencing conversion and churn prediction.
*   Insights and Realtime Dashboards with Tableau: Used Regression, Classification and Clustering algorithms and monitored KPIs, trends, and anomalies based on guidance and improved Conversion Rate by 10%, Return on Marketing Investment(ROMI) by 6%, and Net Promoter Score by 3 points.

  

*   \[Discover Bank\] - Multi-Label Text Classification and Summarization using Attention-based Transformer models like BERT and T5
*   Deep Learning models: Implemented RNN and LSTM classifiers using ELMO embeddings and fine-tuned pre-trained BERT models resulting improved F1 score to 93%.
*   Legal Summaries using BERT and T5: Leveraged Encoder-Decoder and Attention based Transformer models to summarize litigation documents by fine-tuning pre-trained models like BERTSUM, and Text-to-Text Transformer (T5) models.

Data Scientist II

1.  \[HBTF Bank\] - Predictive Analytics for augmenting Customer Experience for banking product (Finacle)

*   Anomaly Detection to reduce Financial Fraud: Overcame cold start problems by employing K-means clustering for preliminary anomaly categorization. Enhanced this framework with the Isolation Forest algorithm, significantly boosting fraudulent transaction detection accuracy to 92%.
*   ETL Pipelines using PySpark and Databricks: Optimized ETL pipelines using PySpark and Databricks, enabling efficient data ingestion into cloud-based storage AWS S3. Enhanced real-time and batch data processing, significantly improving dynamic dashboard updates and data reliability.

Data Scientist

1.  \[Sohar Bank\] - Optimizing Data Integration for real-time Analytics and Reporting
2.  • SQL Query Optimization with Advanced Techniques: Employed advanced SQL optimization strategies including indexing, pruning and integrated with HDFS and PySpark workflows to ensure swift data retrieval and reduce server loads, leading to a 24% improvement in database performance.

Bindu Latha Banisetti

#1 Video-Talent Marketplace

Ready to discover your perfect career?