Headline

𝘐𝘵𝘦𝘳𝘢𝘵𝘪𝘯𝘨…

About

What's interesting about machine learning is the gap between models that look smart and models that actually are. The hard cases are the small ones — small deviations from a pattern, in a system that's been watching long enough to know what "normal" looks like for this specific user, this device, this time and teaching a model to notice that — and flag what's unusual without crying wolf at everything new — is much harder than it sounds. Closing that gap - building things that handle edge cases - messy data - distributions they weren't trained on feels like the actual frontier — that's the kind of work that holds my attention.

Location

Bengaluru

AI stack

claude code

claude

hugging face

kiro

notion

Vercel AI SDK

Work experience

Intern at F-Secure Corporation, Feb 2026 - Jul 2026 · 6 mos, Built Care-Agent: Node.js cloud automation platform managing 100s of AWS accounts via STS AssumeRole. Implemented parallel async log retrieval using CloudWatch Logs Insights reducing time from hours to seconds. Developed AWS MCP tool automating SSO authentication and CloudWatch log fetching pipeline, cross-referencing CloudWatch and Dynatrace logs to troubleshoot the services throwing errors.

Machine Learning Intern at Prodigy InfoTech, Jun 2025 - Jul 2025 · 2 mos, Completed ML tasks: Task 1 House Price Prediction (Linear Regression). Task 2 Customer Segmentation (K-Means). Task 3 Cat-vs-Dog Classifier (SVM + HOG, 86% accuracy). Task 4 Hand Gesture Recognition (CNN on LeapGestRecog, 10 classes).

Education

Master of Computer Applications, Data Science, CMR Institute Of Technology

Bachelor of Computer Application, New Horizon College of Engineering

Public links

github: https://github.com/Vagvane

leetcode: https://leetcode.com/Vagvane