ad4tnu@r.postjobfree.com; ***-SHIVAANSH ***-***0; linkedin.AGRAWAL com/in/shivaansh-agrawal; github.com/shivaanshag; EDUCATION

Master of Science, Computer Science (AI & ML Specialization) May 2024 The University of Texas at Dallas, Texas GPA 3.97/4 Relevant Coursework: Computer Algorithms, Data Structures, Database Design, Big Data, ML, AI, NLP, Web Tech Bachelor of Technology, Electronics and Telecommunication Engineering May 2019 National Institute of Technology, Raipur, India (Awards: Honors, PMs Trophy Scholoarship) GPA 8.25/10 Relevant Coursework: Computer Architecture, Microprocessors, Operating Systems, Information Theory TECHNICAL SKILLS

Programming : C/C++, Python, Java, SQL, HTML, JavaScript, Bash Certifications : Deep Learning with Python, Video Technology (Ittiam Systems Pvt Ltd), Software Architecture Foundations Softwares : TensorFlow, GNU Debugger (GDB), scikit-learn, PyTorch, pandas, Spark, Hadoop, Docker, AWS, matplotlib, OpenCV Softwares : Jira, Postman, AirFlow, Kafka

ACADEMIC PROJECTS

Accelerated Image Filtering, CS 6384 Computer Vision, UT Dallas, TX, USA CUDA C++, OpenCV Spring 2024

• Implemented GPU accelerated image filtering algorithms: linear, gaussian, bilateral, non-local means using CUDA C++ and OpenCV for image operations to attain massive speedup vs CPU implementation. Distribited ML Training on Cloud, CS 6350 Big Data, UT Dallas, TX, USA AWS, numpy, Apache Spark Fall 2023

• Trained RNN and LSTM prediction model on stock market data on distributed AWS cluster and achieved 10x speedup vs local. Nine-Men’s Morris Game Agent, CS 6364 AI, UT Dallas, TX, USA AI Game Trees, Python Summer 2023

• Developed an innovative gameplay agent for the modified Nine-Men’s Morris game using Minimax & Alpha Beta Pruning algorithm on adversarial game trees. Achieved 2nd rank (out of 45) in inter-class game agent tournament. Class-conditional GANs & VAE, CS 6375 ML, UT Dallas, TX, USA Python, Pytorch, CUDA Spring 2023

• Designed an image Generative Adversarial Network (GAN) for the MNIST & 80 Million MIT tiny-images dataset.

• Architecture achieved generation of class-conditional images; the discriminator could also output class labels.

• Implemented a Variational Autoencoder (VAE) for the same datasets with varying latent dimension sizes. Computed the best compression latent dimension with minimal information loss. TransGANs, ET20823 Major Project, NIT Raipur, India Python, TensorFlow Spring 2019

• Engineered a GAN for Sanskrit Shloka generation. Conceptualized a novel Transformer GAN architecture based only on attention mechanisms which does not use convolutions or recurrence achieving parallelism and no context loss.

• TransGANs training computations are < 0.1 times that a comparable model (SeqGAN) on avg sequence length of 42. EXPERIENCE

Senior Software Engineer, Ittiam System Pvt Ltd, Bengaluru, India Oct 2021 – Jun 2022

• Led a team of 5 to validate major audio framework services for Android 13 release. Project delivered with 4 large-size personal contributions to the Android Open Source Project (AOSP).

• Devised multi-threaded JUnit tests for Media Performance Class in Android 13, resulting in 20 git contributions to AOSP and ensuring consistent media experience for billions of users.

• Developed and validated multimedia components in Android OS on Java & native C++ layer as a Google partner (Google Center of Excellence), contributing to 8 Statements of Work (SoWs) from Google Android Media Team.

• Added enhancement in H.264 (AVC) & H.265 (HEVC) video decoder in C language to export info-map and integrated it to the codec framework by modifying the codec plugins delivering 8 CLs (Change Logs)/contributions. Software Engineer, Ittiam System Pvt Ltd, Bengaluru, India Jul 2019 – Oct 2021

• Debugged and resolved four high-security & 12 low-security risk bugs/issues for the Android Media framework’s S/W codecs & media pipeline media components; engineered 3 CLs for Codec Validation tests.

• Implemented 6 CLs to enhance test coverage of media components using C++ unit tests by 30% (from 50% to >80%).

• Designed Media Benchmark Android application (Java & C++) to assess media performance across devices, Android versions(4+), and codecs (10+: AV1, HEVC, AVC, VP8, VP9, MP3, AAC, ...). Software Intern, Divergent Software Labs Pvt Ltd, Indore, India May 2018 – Jul 2018

• Researched SNS feed & geospatial recommendation algorithms to develop ML-based user feed algorithm (using PredictionIO) for Coloneed, a hyper-local SNS with location-based posts, leading to 20% increase in user engagement. LEADERSHIP / EXTRACURRICULAR

Core Member, Entrepreneurship Cell, NIT Raipur, India Aug 2017 – May 2019

• Created a Wall Street simulator using actual data played by 100+ players for a stock market competition.

• Coordinated sponsorship events; VCs invested more than INR 12 million ( $100k) across all teams.