AI Developer Intern @ Stanza AI (L-VelUp Edulab FZCO) | GenAI & RAG Engineer | Final-Year B.Tech CSE (AI&ML)
I'm a final-year B.Tech CSE (AI & ML) student (CGPA: 8.55/10) building production-grade GenAI, RAG, and multi-agent systems. Currently working as an AI Developer Intern on Stanza AI, an AI layer for a hospitality operations platform — fine-tuning proprietary LLMs (LoRA/QLoRA) to cut dependency on external APIs, across dataset strategy, evaluation, and agent orchestration.
- 🔭 Building agentic RAG pipelines with LangGraph, LangChain, and FAISS
- 🧠 Fine-tuning LLMs with LoRA/QLoRA, exploring proprietary model development
- 🌱 Currently deep in LangGraph (MemorySaver, thread/session state, streaming) via a Gen AI course
- 🏆 Bronze Medalist, IIC 2025 | GSSoC 2025 & 2026 Contributor
- 📫 Reach me at raviroy2002@gmail.com
An end-to-end Agentic RAG system over Indian legal acts (POCSO, RTI, Consumer Protection Act 2019) — zero manual curation, fully automated ingestion to deployment.
- ⚙️ LangGraph agentic RAG with tool-based FAISS retrieval (top-4 chunks),
MemorySaverconversational memory, and automatic topic routing across 3 legal acts - 📚 309 FAISS vectors from HuggingFace
all-MiniLM-L6-v2embeddings over cleaned, chunked legal PDFs - 🤖 3 automated output modes via Groq LLaMA 3.3-70B: 250-word summaries, structured JSON key-info extraction, and gTTS audio generation
- 🐛 Resolved 5 production bugs (encoding, JSON fencing, FAISS metadata filtering, Groq tool-call/streaming conflicts, session config injection)
- 🔗 GitHub · Live Demo
| Project | Description | Links |
|---|---|---|
| Sentiment Analysis (RNN/LSTM) | NLP pipeline on 50K IMDB reviews, ~76% test accuracy | GitHub · Live |
| Customer Churn Prediction (ANN) | Deep learning model on 10K+ records, 86% accuracy, TensorBoard-monitored | GitHub · Live |
| Customer Segmentation | Segmentation dashboard deployed on Render | GitHub · Live |
| Heart Disease Prediction | Classification model deployed on Render | GitHub · Live |
| Credit Card Fraud Detection | XGBoost + SMOTE on 284K+ transactions, 91% recall on imbalanced data (not deployed) | GitHub |
GenAI & RAG
ML / DL
Deployment & Tools
Languages: Python · SQL · Java · C
- B.Tech, CSE (AI & ML) — IILM University, Greater Noida (2023 – Jul 2027), CGPA: 8.55/10
- ML Specialization — DeepLearning.AI
- Apply AI & Modern AI — Cisco
- Responsible AI, LLMs, GenAI — Google Cloud
Open to GenAI / ML Engineering roles — let's connect!