Building AI Systems, Full-Stack Applications & Open Source Projects

• 🎓 B.Tech CSE (AI Major), Minor: FinTech — RV University, Bangalore | CGPA: 9.19/10
• 🤖 Focused on Reinforcement Learning, LLM fine-tuning, and full-stack AI systems
• 🏆 Top 2.5% of 32,000+ teams — Meta PyTorch OpenEnv Hackathon
• 🌱 Active Open Source Contributor (GSSoC 2026, C2SI Organisation)

• 🏆 Ranked Top 800 / 32,000+ teams at Meta × Scalar All India AI Hackathon
• 🤖 Fine-tuned Qwen2.5-3B via GRPO — full-fix rate 0% → 91% at peak batch
• 🏗️ Shipped CampusRunner: full-stack monorepo with real-time delivery, payments, and CI/CD
• 📊 Designed E-Commerce DBMS with normalized schema, REST API, and 8+ entities
• 🔬 Built adversarial two-agent RL environment for compliance patch training

• 🤝 Contributing to GSSoC 2026 and C2SI open source projects
• 🤖 Adversarial LLM fine-tuning and RL-based compliance systems
• 🕸️ Full-stack platforms with React, Flask, Docker, and CI/CD pipelines
• 📚 Exploring RAG architectures and agentic AI workflows

• Retrieval-Augmented Generation (RAG)
• LLM Fine-tuning (QLoRA, GRPO, Unsloth)
• Backend Architecture with Flask & Node.js
• Docker & Deployment Workflows

Languages

AI / ML & Frameworks

DevOps, DB & Cloud

Full-stack monorepo (React 18 + TypeScript + Flask + SQLAlchemy) with 3 user roles, real-time order tracking via Socket.IO, Razorpay payments, and a full Docker + GitHub Actions CI/CD pipeline with CodeQL and Trivy scanning.

Two-agent adversarial RL environment (Patcher vs. Adversary). Fine-tuned Qwen2.5-3B-Instruct via GRPO (TRL + Unsloth 4-bit QLoRA) — reward 0.84 → 2.11, full-fix rate 0% → 91%. Deployed on HuggingFace Spaces + Docker.

PyTorch RL environment for auditing GDPR, SOC2, HIPAA, PCI-DSS, and ISO 27001 compliance. AST-level semantic patch validation across a 3-tier difficulty ladder. Top 2.5% at Meta PyTorch OpenEnv Hackathon.

Self-validating autonomous agent that re-evaluates each action against expected state before committing. Selected for Round 2 in a 24-hour competition.

Flask + MySQL e-commerce system with authentication, product catalog, cart, order management, normalized DB schema, and REST API layer.

AI Intern — Broadrange AI (Jan – Jun 2024) Built a production Streamlit multimodal app integrating Hugging Face APIs (BART, RoBERTa) for summarization, Q&A, translation, and TTS. Engineered PDF extraction, audio transcription, and token-cost optimization as sole developer. → github.com/RKBroadrangeAI/rvmultimodal

2026
├── Joined GSSoC 2026
├── Contributing to C2SI: acf-sdk, dataloom, honeynet
└── Learning large-scale open-source collaboration

• Open Source Projects (AI, ML, Backend)
• RL & LLM Applications
• Full-Stack Development
• Backend Engineering

I enjoy turning academic concepts like queues, compliance frameworks, and NLP into systems that actually run in production.