VRB Capital LLC
AI Automations & IT Systems Intern · Remote
- Built Claude-powered automation pipelines for internal operations.
- Integrated Zoho Bigin CRM with custom Python workflows.
- Maintained cPanel infrastructure and ChatGPT-assisted tooling.
AI · ML · Inference in motion
I work at the intersection ofresearch, engineering,and design.From RAG agentsto federated learning,every system I ship is builtto be fast, honest, and useful.

I'm an AI / ML engineer studying at BMS Institute of Technology, where I'm pursuing a B.E. in Artificial Intelligence & Machine Learning (CGPA 8.91).
I build production-grade AI systems — RAG agents, federated learning models, deepfake detectors, recommendation engines — and pair them with full-stack interfaces that respect the user's time. My research focuses on healthcare AI and privacy-preserving ML.
Currently interning at VRB Capital LLC — building Claude-powered automation pipelines for internal operations.
AI Automations & IT Systems Intern · Remote
Healthcare AI · Federated Learning · Bengaluru, IN
10 RAG architectures compared side-by-side — Hybrid, Graph, Agentic, Corrective, Multimodal, Multilingual, RAG-Fusion, HyDE, Structured, and Self-RAG. Upload docs, ask once, see all pipelines answer in real time with RAGAS eval scores.
ATS-grade AI agent that scores resumes via 50% keyword + 50% FAISS vector similarity, with a self-learning RAG pipeline for rewrites, cover letters, and interview prep.
HIPAA/GDPR-aligned federated learning system for clinical sepsis detection — AUROC 0.894, AUPRC 0.567. Live demo + peer-reviewed research paper.
Vision Transformer + ConvNeXt ensemble with Grad-CAM explainability. AUC 0.9977 over 16,000 face crops — production FastAPI service.
MusicGen + local Llama-2 agent generates and orchestrates compositions. Next.js 14 frontend, FastAPI backend, on-device inference.
Personal finance app with budgeting, expense tracking, and analytics. React + TypeScript + Supabase, packaged for mobile with Capacitor.
LLM-generated itineraries with interactive maps and supabase-backed trip storage. Built with TypeScript, Gemini & Groq.
XGBoost-powered sales & demand forecasting with feature engineering on retail timeseries — live Streamlit dashboard.
NLP-driven ticket classification with DistilBERT — ~90%+ accuracy on multi-class routing. Live Streamlit demo for triage teams.
Cross-platform task manager with Spring Boot backend, React web client, and Flutter mobile app — shared API surface.