Experience
Undergraduate Research Assistant
UW–Madison Bhaskar Lab | OpScribeAI
May 2026 - Present
I work on developing and evaluating large vision-language models (VLMs) for surgical video understanding and AI-assisted surgical documentation. My work focuses on improving how multimodal models interpret surgical procedures, including identifying instruments, anatomy, surgical phases, and procedural steps.
I build data preprocessing pipelines in Python to transform raw surgical video annotations into structured VQA datasets for model training and evaluation. One of the datasets I worked with contains 8,800+ question-answer pairs spanning instrument, anatomy, phase, and step categories.
I also run QLoRA fine-tuning experiments on 72B-parameter VLMs using PyTorch and Hugging Face Transformers. I evaluate model performance across different surgical reasoning tasks and compare results against other vision-language models to understand where different approaches perform well and where they struggle.
Through this work, I’ve gained experience working with large-scale multimodal models, GPU/HPC computing environments, model fine-tuning, dataset development, and evaluation of AI systems in a specialized domain.
Cybersecurity Intern
Kohler Co.
May 2026 – Aug. 2026
I work on the cybersecurity team, focusing on applying AI and automation to enterprise security operations. A major part of my work involves building AI security agents in Microsoft Copilot Studio to automate tasks that would otherwise require significant manual investigation and review.
I’ve developed several AI-driven security tools, including an architecture review assistant, IoT360 security assessment assistant, and AI threat hunter. These agents use organizational knowledge and security data to assist with architecture reviews, security assessments, and threat investigations. The tools achieved 100% knowledge utilization, increased the knowledge answer rate by 33%, and reduced manual review effort by 60–80%, while enabling automated SOC investigations around the clock.
I also design and test workflows connecting security platforms including Cortex XSIAM and ServiceNow. This involves validating API interactions, backend data flows, authentication, and automation logic to ensure that information moves reliably between different systems.
In addition, I support SSL/TLS certificate automation by integrating Sectigo with Microsoft IIS. I’ve helped validate automated certificate issuance and renewal workflows covering 141+ certificates, reducing the risk of certificates expiring and disrupting enterprise services.
This role has given me experience at the intersection of AI, software engineering, cybersecurity, and enterprise automation, while exposing me to how AI systems can be integrated into real-world security operations.
Website Lead & Finance Lead
Theta Tau – Xi Chapter, UW–Madison
Dec. 2024 – Present
As Website Lead, I design, develop, and maintain the chapter’s official website. I work across the site’s frontend and backend functionality to keep information up to date and make the website easier for members, prospective members, alumni, and visitors to navigate.
One of my larger projects was implementing a members page for 300+ members, creating a more organized way to showcase the chapter and its community. I’ve also worked on improving the site’s functionality and resolving page issues, contributing to a 90% reduction in page load errors and a 1000% increase in site traffic.
The website was recognized with Best Website among 94+ Theta Tau chapters nationwide, which was especially meaningful because the award recognized both functionality and user experience.
As Finance Lead, I manage the chapter’s finances and budgeting for 100+ members. I track member dues, event expenses, and semester allocations while working to keep spending organized and financially accountable.
Balancing both technical and financial responsibilities has given me experience managing a project from both the implementation and organizational sides.
Tech Consultant
UW–Madison Department of Information Technology (DoIT)
Apr. 2025 – Sep. 2025
As a Tech Consultant, I provided technical support to 50,000+ UW–Madison students, faculty, and staff through the university’s IT support services.
I handled 30+ service requests per day involving hardware, software, account access, networking, and other technology issues. I worked directly with users to understand their problems, troubleshoot potential causes, and identify solutions while communicating technical information in a way that was accessible to people with different levels of technical experience.
The role also required me to manage multiple requests efficiently, document troubleshooting steps, and escalate issues when they required specialized support.
Working in a large-scale university IT environment strengthened my troubleshooting, communication, and problem-solving skills and gave me experience supporting technology systems used by a large and diverse community.