Research Intern

TINENDRA KANDULA

+1-617-***-****, Boston, MA ********.*****@*****.*** linkedin.com/tinendra-kumar EDUCATION

Master of Science in Computer Science Sept 2022 - May 2024 University of Massachusetts Lowell (UML) Lowell, MA Relevant Coursework: Machine Learning, Algorithms, Design of programming languages, Advanced Database systems Master of Technology in Communication Systems 2021 - 2022 SASTRA University Thanjavur, India

Thesis: Modulation Recognition in Cognitive Radio Networks using Deep Learning. Bachelor of Technology in Electronics and Communication Engineering 2017 - 2021 SASTRA University Thanjavur, India

Thesis: Early Parkinson’s diagnosis using Machine Learning SKILLS

Programming: Python, Javascript, SQL, C, HTML/CSS, Matlab, Embedded systems Tools/Libraries: PyTorch, Transformers, TensorFlow, Scikit-learn, SLURM, Docker, Git, Vue.js, AWS, OpenCV RESEARCH EXPERIENCE

Research Intern - Shafiee Lab Oct 2023 - Present

Harvard Medical School Cambridge, MA

• Developed a Multi-Modal deep learning system for IVF live birth prediction that combines parsed clinical notes using BioBERT model, embryo videos through multi-instance learning, and clinical tabular characteristics using a cross-modal transformers-based fusion framework. This multimodal system enhanced predictive F1 scores to 76% compared to 68% with unimodal approaches. (Submitted to the ASRM conference)

• Implemented advanced deep learning models and conducted statistical analyses using three bias mitigation strategies for annotation bias, instrumentation bias, and model variability in live birth prediction and embryo grading. (Submitted to Nature BME journal)

• Configured and managed the lab’s multi-node distributed GPU cluster with SLURM job management for scalable machine learning tasks, including setting up a high-speed network and shared NFS storage. Research Assistant - Communication Systems Lab Oct 2021 - July 2022 SASTRA University Thanjavur, India

• Developed and implemented Convolutional Neural Networks (CNNs) to enable effective digital modulation switching in high-frequency communication systems like 5G.

• Achieved an accuracy of 95% with the CNN model, improving power efficiency without compromising complexity. PUBLICATIONS

• Jiang, V.S., Kanakasabapathy, M., Thirumalaraju, P., Kovilakath, N., Kandula, T., et al. (2024). Enhancing IVF success: Integrating Patient Data, Cycle Metrics, and Embryo Imaging (Submitted to ASRM Conference)

• Thirumalaraju, P, Kanakasabapathy, M., Kandula, T., Souter, I., et al. (2024). Implementation Challenges and Bias Mitigation in Embryology (Submitted to Nature BME Journal) PROJECTS

Interactive UI/UX WebApp for Car Decision Platform Advisor: Dr. Levkowitz (UML)

• Designed and implemented a user-friendly website using Vue.js, HTML, and CSS providing an interactive plat- form for users to compare the financial aspects of buying versus leasing a car. Ensemble Models for Gas Detection Advisor: Dr. Ruizhe Ma (UML)

• Investigated the effectiveness of logistic regression outputs integrated through a voting ensemble for classification, incorporating time series analysis of sensor data, and found that a standalone MLP model surpassed ensemble methods, achieving 85% accuracy in distinguishing gas types. Smart Stick for Visually Impaired People. Robotics Club (SASTRA)

• Built a smart stick for visually impaired individuals for indoor navigation, incorporating object detection algo- rithms (TensorFlow, OpenCV, Python) to detect obstacles in real-time from a video feed.

Contact this candidate