|
|
|
I am a graduate student at the University of California, San Diego, pursuing a master's in Electrical & Computer Engineering with a specialization in Machine Learning and Data Science. My academic journey is deeply rooted in exploring the intersection of distributed systems and advanced machine learning techniques. Prior to my graduate studies, I gained over two years of professional experience as a Software Engineer at Arista Networks and Pluribus Networks. In these roles, I worked on enhancing software-defined networking (SDN) solutions, focusing on features integrating kubernetes and openstack in netvisor OS. I led initiatives to develop and implement automation frameworks that ensured high-performance and resilient networking solutions, directly contributing to operational efficiency and customer satisfaction. Beyond my technical endeavors, I am enjoy contributing to the tech community through open-source projects and collaborations with fellow researchers and professionals. My enthusiasm extends beyond technology, as I also cherish exploring nature through hiking and trekking. I am keen to further my expertise in distributed systems and machine learning and am always open to new conversations. Feel free to check out my Resume and drop me an e-mail if you want to chat with me! ~ News | Experience | Projects | Articles | Contact ~ |
|
May '24 |
Presented my research poster and demo at San Diego Wireless Summit 2024. |
|
Feb '24 |
Started working as a Machine Learning Graduate Researcher at Center for wireless communications(UCSD). |
|
Sept '23 |
Started Fall 2023 Quarter - U.C. San Diego! |
|
July '23 |
Completed my 1 years working as a full-time Software engineer at Arista networks. |
|
July '22 |
Completed my 1 years working as a full-time SDE at Pluribus networks. |
|
Feb '24 - Present |
|
July '22 - Sep '23 Led a 7-member team to create a Python framework for Netvisor OS, streamlining feature verification tailored to customer needs. Architected solutions for a leaf-spine distributed network, optimizing both north-south and east-west traffic flows. Enhanced system efficiency by reducing regression verification time by 40% through multithreading and anomaly detection. |
|
July '21 - June '22 Developed infrastructure to validate and monitor Kubernetes network operations integrated with Netvisor OS, enhancing visibility and diagnostics. Deployed Kubernetes on ESXi VMs and implemented network path tracing between pods. |
|
|
Jan '21 - June '21
Streamlined VRRP Cluster Router Certificate Upgrades: |
|
Oct '20 - May '21 Developed an AI Application for Radiologists: |
|
University of California - San DiegoMajor: Machine Learning & Data Science Sep '23 - Present Relevant Coursework: Statistical Learning • Recommender Systems and Data Mining • Machine learning for physical applications • Linear Algebra •Probability and Statistics for Data science • Introduction to Visual Learning |
|
PES University, BengaluruSpecialization: Communication systems Aug '17 - May '21 |
|
Pytorch, Hugging Face , OpenCLIP | April. 2024 - June. 2024 [code]
Developed a zero-shot image captioning framework that significantly enhances model performance by fusing the k nearest neighboring CLIP text embeddings with BLIP image embeddings using a Graph Convolutional Network (GCN). This enhancement was achieved by establishing a knowledge database containing CLIP image and text embeddings, from which we extracted the k closest image-text pairs based on CLIP image embeddings. The Q-Former was fine-tuned in conjunction with the GCN-fused architecture, resulting in captioning capabilities that surpass current state-of-the-art (SOTA) models. |
|
Pytorch, dm-control, torchdiffeq, gym simulation | April. 2024 - June. 2024 [code] Modelled an ODE-Enhanced Neural Network Model to understand the dynamics of systems with high degrees of freedom, such as walkers, specifically tailored to address the complexities of movements derived from irregularly sampled Mujoco dynamics. Utilizing the DYNA-Q pre-trained walker model, we generated accurate ground truth trajectories, which enabled our model to adeptly predict and analyze dynamic behaviors across varying data sampling intervals. |
|
Pandas, Numpy, Plotly, Jupyter Notebook | Nov. 2023 - Dec. 2023 Analyzed over 1 million reviews in the Alaska Google Local Dataset, understanding user behaviors and preferences to forecast ratings.Predicted rating using various predictive models, including latent factor models, factorization machines, and linear regression. Achieved a 0.2605 reduction in MSE over the baseline by incorporating detailed text analysis and category insights into linear regression. |
|
Keras, Tensorflow, Copula | Jun 2020 - Dec 2020 [code] Developed a statistical model using R-Vine copula for efficient multimodal data representation and dimensionality reduction. Surpassed the state-of-the-art by 10.24% in activity recognition with phone and watch sensor data for STISEN dataset. |
|
Keras, Tensorflow, OpenVINO , Jupyter Notebook | Jan. 2020 - Apr. 2020 [code] Designed and trained a Keras model to locate existing infarcts within 1000+ raw MRI images with an accuracy of 86.7%. Enhanced the limited dataset using data augmentation techniques; Deployed the model for inference using Intel OpenVINO |
|
Jun '24 |
Explored the scenic Santa Margarita trail, covering 6.4 miles. |
|
Mar '24 |
Explored an undefined 4-mile trail in Mission Valley , San Diego |
|
Jan '24 |
Conquered the 5.2-mile Iron Mountain trail in Poway, California, reaching an elevation of 2,696 feet. |
This template is a modification to Jon Barron's website. Find the source code to my website here.