Contact Information

Email: mrinaltyagi24@gmail.com

LinkedIn: Mrinal Tyagi

Medium: mrinaltyagi24

GitHub: MrinalTyagi

Education

  • Master of Science in Research (CSE)
    IIT Delhi, New Delhi, India (2024 - Present)
  • Bachelor of Technology (ECE)
    Bharati Vidyapeeth’s College of Engineering, New Delhi, India (2019 - 2023) - GPA: 8.41

Skills Summary

Languages: C, C++, Python, Swift, MySQL

Tools: Git, XCode, Jupyter Notebook, VSCode, Sublime, AWS

Libraries/Frameworks: PyTorch, Tensorflow, Plotly, Scikit Learn, OpenCV, Pandas, PyQt5, MongoDB, Flask

Key Concepts: Problem Solving, Data Structures & Algorithms, Statistics, Probability, Linear Algebra, Operating System

Experience

Project Scientist

IIT Delhi, New Delhi, India (Feb 2024 - Present)

Working on video instance segmentation for surgical videos.

Advised by: Prof. Chetan Arora
Co-advised by: Prof. Prem Kumar Kalra

Jr. Project Assistant

IIT Delhi, New Delhi, India (Nov 2023 - Feb 2023)

Developed a new flow consistent augmentation technique for temporal action segmentation, accepted at MICCAI 2024.

Advised by: Rohan Raju Dhanakshirur, Prof. Chetan Arora
Co-advised by: Prof. Prem Kumar Kalra

Research Intern

IIT Delhi, New Delhi, India (April 2023 - Oct 2023)

Developed suturing evaluation software for medical institutes.

SDE Intern

Zomato, Gurugram, India (Jan 2022 - July 2022)

Worked on the Dining Module for iOS, enhancing UX and SEO features, resulting in a 4.8% increase in revenue.

Data Science Intern

Innomatics Research Labs, Remote (April 2021 - July 2021)

Worked on statistical modeling, hypothesis testing, and a music recommendation system.

Publications and Conferences

  • Custom Weighted Balanced Loss Function for COVID-19 Detection from an Imbalanced Dataset
    26th International Conference on Pattern Recognition, Montreal, Canada
    Manuscript | GitHub
  • SVD-CLAHE Boosting and Balanced Loss Function for COVID-19 Detection from an Imbalanced Chest X-Ray Dataset
    Computers in Biology and Medicine (Elsevier Q1)
    Manuscript | GitHub
  • A Transformer-based Approach for Abuse Detection in Code Mixed Indic Languages
    ACM Transactions on Asian and Low Resource Language Information Processing
    Manuscript

Projects

  • COVID-19 Vaccine Prioritizer using Clustering - GitHub
    Developed a data-driven tool using KMeans clustering to prioritize COVID-19 vaccine distribution across Indian districts.
  • Neural Network from Scratch in Swift - GitHub
    An interactive playground for beginners explaining neural networks using Swift.
  • Other Projects
    Employed HAAR Cascade, Single Shot Detector (SSD), and Segmentation models on multiple datasets.

Open Source Contributions

  • TensorFlow: Added a new hyperparameter beta in the swish activation function.
  • Scikit-Learn: Code cleanup for various sections including GridSearchCV, HalvingSearchCV, and other test fixes.
  • Hugging Face: Code optimization in forward pass of inference.
  • Keras CV: Documentation cleanup and addition of YOLOv7 in the applications section.

Extra Activities

  • Kaggle Expert:
    Datasets Expert (Rank 71/71,194), Notebook Expert (Rank 264/238,257), Discussion Expert (Rank 500/308,227)
  • Medium Writer: Published articles on topics such as HOG (Histogram of Oriented Gradients), Viola Jones Algorithm, Image Segmentation, and more.