• I am an experienced Software Engineer and Technical Leader working in Personalization and Recommender Systems Space.
• Top Skills: Java, Python, Hadoop, Hive, Cassandra, Spark, Spring, Kubernetes, TensorFlow, GCP, A/B Testing, Personalization Algorithms, Data Pipelines
• 15+ years of experience delivering innovative solutions in large-scale distributed systems, big data, and machine learning environments.
• Proven track record of leading engineering teams, optimizing system performance, and driving significant business impact.

Below is a list of recent academic projects I have worked on.

• Description: Developed a computer vision model for detecting glaucoma from fundus photographs using Vision Transformers, achieving 90% accuracy.
• In collaboration with Amit Arora, Arindam Sett and Malay Patel
• Accomplishments:
  • Implemented an ensemble approach combining multiple ML algorithms (e.g., CNNs, SVMs) to improve model robustness.
  • Collaborated with a team of data science students, ophthalmologists, and an academic advisor.
  • Utilized transfer learning and fine-tuning techniques on a limited dataset.

• Description: Developed a Question Answering model to extract answers from product reviews, focusing on subjective questions.
• In collaboration with Malay Patel
• Accomplishments:
  • Leveraged transfer learning and cross-pollination techniques to improve model performance.
  • Experimented with various pre-trained models (BERT, DistilBERT) and datasets.
  • Performed error analysis to identify areas for improvement.

• Description: Developed and deployed deep learning models for vehicle and license plate detection using YOLOv5 and EasyOCR.
• In collaboration with Arindam Sett and Malay Patel
• Accomplishments:
  • Utilized cloud-based GPU instances on AWS for training and optimization.
  • Evaluated model performance using [email protected] and [email protected], focusing on recall.
  • Integrated multiple pre-trained models with DeepStream SDK and TensorRT for real-time detection.

• Description: Developed a solution for predicting flight delays 2 hours before departure.
• In collaboration with Kumar Narayanan, Ajeya Jayaram
• Accomplishments:
  • Performed exploratory data analysis (EDA) on large datasets (630M records).
  • Created and evaluated various machine learning models (logistic regression, random forest, MLP).
  • Implemented feature selection techniques (PCA, one-hot encoding).

• Description: Developed a Robo Advisor for Loan Applications using machine learning.
• In collaboration with Hailemariam Bizunehe and Kris Junghee Lee
• Accomplishments:
  • Performed EDA on a dataset with over 2 million records and 151 columns.
  • Implemented feature engineering techniques (handling missing values, encoding, text preprocessing).
  • Addressed imbalanced data using resampling techniques (SMOTE, ADASYN).
  • Evaluated various models (logistic regression, random forests, XGBoost), achieving a test F1 score of 0.91.