Contact
Pittsburgh, PA
(+1) 412 909 9156
Email • Website • LinkedIn • Github
Education
Carnegie Mellon University • Pittsburgh • PA
Master of Science • Artificial Intelligence and Innovation
Anticipated graduation - 05/26
GPA: 3.93
College of Engineering Trivandrum • Trivandrum • India
Bachelor of Technology • Computer Science & Engineering
Graduated - 07/21
GPA: 9.0
Skills
Python, PyTorch, Java, JavaScript, C++, PyTorch, SQL, React, Redux, Next.js, TypeScript, Azure, HTML, CSS, Mixpanel
Work Experience
Providence Global LLP – Software Engineer 2
August 2021 – July 2024
Hyderabad, India
-
Designed and implemented client-side analytics for trusana.com with A/B testing, Mixpanel integration, and user cohort analysis, driving a 9% boost in conversion rates over 4 months.
-
Optimized website load time from 12 seconds to P99 under 3 seconds using GZIP compression, webpack bundle optimization, and React lazy loading, reducing user dropoff rates by 40%
-
Built a pipeline for de-identification and summarizing/generating patient responses/charting notes using fine tuned local models (LLaMa-7B)
-
Delivered a Trusana landing page within 2 weeks, ensuring WCAG 2.0 AA compliance with tools like ARIA landmarks and keyboard navigation, screen reader support.
-
Led a team of 3 engineers to develop a platform from scratch to support hospital front-line workers, reducing workload-related burnout incidents by 15%.
-
Delivered a healthcare platform from scratch to MVP within 2 months and to full production in another 3 months, providing real-time support for over 500 caregivers.
-
Promoted in 1.5 years for leading delivery of key MVPs, including an E2E video calling feature and an iOS app, showcasing rapid learning and cross-functional impact.
Projects
LoRA implementation
November 2024
Technologies: Python, PyTorch
-
This notebook was a small project to understand LoRA finetuning. It implements LoRA from scratch primarily using the paper as a guide.
-
Achieved 97.9% of the performance of normal finetuning with as little as 7.7% of the trainable weights compared to the traditional approach.
Accessibility - enabled by LVM
October 2024
Technologies: Python, PyTorch
-
Designing an accessibility enhancement app that leverages large vision models (e.g., GPT-4V) to provide screen-reader compatibility and voice-controlled navigation for websites lacking built-in accessibility.
-
Employing Claude APIs and GPT-4 voice models to optimize real-time user interaction workflows for accessibility, focusing on seamless action execution and user adaptability.
RAG pipeline for library specific docs
September 2024
Technologies: Python, PyTorch, Pinecone
-
Developed a RAG based interface for library documentation which doesn’t exist in training data for gpt models.
-
Scraped docs, converted them to md with firecrawl, chunked the data and setup a pipeline for MIPS to surface relevant chunks, on pinecone.
Image Diffusion Model
August 2024
Technologies: Python, PyTorch
- Built a diffusion model using a custom UNet architecture and PyTorch to create denoised abstract images, utilized as thematic visuals for blog posts on my website, showcasing proficiency in generative AI techniques.
Pittsburgh 2 Peers
June 2024
Technologies: React, .NET, Vercel
Live Demo • Recorded Demo
-
Developed a React/.NET-based community app for CMU students, facilitating rideshare and bulk purchasing coordination, achieving 500+ active users within 3 months.
-
Conducted requirements analysis, market research, and UX studies using tools like Figma, leading to the successful deployment of a community engagement platform.
BERT Question-Answering System
August 2024
Technologies: PyTorch, Azure CR
- Developed a BERT QA system fine-tuned on the SQuAD dataset, and deployed it on Azure Container Registry as an API.
Achievements
-
Awarded the IEEE Richard E. Merwin Scholarship, recognizing global excellence in academics and volunteering as one of only 16 recipients worldwide.
-
Won the Nucleate Pittsburgh Biotech Hackathon out of 200 participants by designing a real-time feedback system for ENT surgeons, leveraging machine learning to enhance cochlear implant success rates.
-
Runner Up, Reboot Kerala | Healthcare Category, out of 30+ State level finalists, for designing a private WebRTC based E2E therapy scheduling solution.
-
Achieved a top 10% ranking (Rank 273 of 3000+) in the IEEE Xtreme 2020 global competitive coding challenge.
-
NTSE Scholar - One of 1000 students selected annually at the national level for academic excellence.
Leadership | Volunteering
- Regional Student Ambassador, IEEE R10 IEEE Computer Society
September 2020 - August 2021 - Student Representative, IEEE Computer Society Kerala Chapter
January 2020 - December 2020 - Chairperson, IEEE Student Branch College of Engineering Trivandrum
February 2020 - January 2021
Last updated: December 26, 2024