1300 N Walnut St, Apt 17, Bloomington, Indiana 47404
+1-812-778-6905 # ashuhath@iu.edu ï linkedin.com/in/ashutosh1919 § github.com/ashutosh1919 Education Indiana University Bloomington August 2021 – May 2023 Master of Science in Computer Science Bloomington, Indiana, USA • Cumulative GPA: 4/4
Indian Institute of Information Technology, Design & Manufacturing July 2016 – May 2020 Bachelor of Technology in Computer Science Kurnool, Andhra Pradesh, India • Cumulative GPA: 9.71/10.0 with Gold Medal
Technical Skills Languages: Python, JavaScript, TypeScript, C++, HTML/CSS, SQL Databases: MongoDB, MySQL, PostgreSQL Deep Learning Libraries: TensorFlow,PyTorch, PyTorch-Geometric, PyTorch-Lightning, spaCy, Scikit-learn Web Technologies: ReactJS, GatsbyJS, NodeJS, D3.js, Plotly Design Tools: Figma, Adobe XD, After Effects, Adobe Premiere Pro Cloud Developer Tools: GCP, AWS, Azure Technologies/Frameworks: Docker, Kubernetes, GitHub Actions, Linux, Jenkins, DVC, CML Experience TikTok Inc. May 2022 – Aug 2022 Software Engineering Intern (Machine Learning) Palo Alto, California, USA • Building new features on the backend recommendation system, specifically ranking algorithms for Ads that touch
hundreds of millions of people around the world.
• Improving online and offline content ranking algorithms by performing hard sample data replays for training steps.
Anthem Inc. June 2020 – July 2021
Associate Software Engineer Hyderabad, Telangana, India • Supervised by Director of AI Dr. Devarkonda Durga Prakash.
• Worked on projects like real time doctor-patient meeting transcription (The Hive), WorkOS for low code development
environment, Medicaid Benefit Search Engine to process insurance claims faster.
• Received Anthem Impact Award 3 times in just 1 year of employment.
Delhivery Pvt. Ltd. May 2019 – September 2019
Applied Data Science Research Intern New Delhi, India • Applied deep learning (LSTM based forecasting), statistical & cloud skills to build pricing engine which forecasts freight
prices. • The freight price prediction is very complex problem because the market is too dynamic. So, we used solution combining
Deep Learning and Statistics to create the pricing engine
Projects Persona2Vec-GCN | PyTorch, PyTorch-Geometric, StellarGraph, Scikit-learn, Numpy, Pip packaging • Objective: To create GCN based representation learning framework for graphs. This is an independent research study
under Dr. YY Ahn at Indiana University.
• Worked closely with Variational Graph Autoencoders to significantly improve the accuracy of neural node embeddings
for link prediction and classification tasks.
Explainable CNNs | Python, PyTorch, PyTorch Lightning, OpenCV, CUDA, Pip Packaging • Objective : To generate layer-wise visual explanations for any PyTorch based CNN classification model.
• Created python pip package to generate different types of layer-wise and class-wise visual explanations based on Grad
CAM, Guided Grad CAM and Saliency maps.
Neuro-Symbolic Sudoku Solver | Python, PyTorch, Jacinle, Reinforce, Pandas, Numpy, Docker • Objective : To solve sudoku using Deep Reinforcement learning in combination with powerful symbolic representations.
• Implemented Reinforcement environment and built codebase to train and infer the solutions of sudoku grid using
Google’s Neural Logic Machines architecture. It solves any sudoku problem with 100% success rate. Face Interpolation and Editing | Python, TensorFlow, Flask, Docker, AWS • Objective : To generate images of faces combining features from multiple face images. (Latent space exploration)
• Created StyleGAN encoder model to create embeddings from the face images and then those embeddings can be linearly
