Hey, I'm Sujit — a B.Tech Computer Engineering student at VIT Pune who builds production-grade AI systems, secures LLMs, and ships real-world cloud infrastructure.
I work at the intersection of AI Security, Cloud-Native Systems, and Generative AI — obsessed with building systems that are not just accurate, but robust, explainable, and scalable.
"Don't just train a model. Deploy one that survives the real world."
🎓 B.Tech Computer Engineering @ VIT Pune | Graduating 2028
📍 Pune, Maharashtra, India
📮 sujitpatil2006@gmail.com
🎯 Open to: Full-Stack · ML/AI · Generative AI Internships
| Title | Venue | Year | |
|---|---|---|---|
| AI-Driven Anomaly Detection for GraphQL APIs under Nested Query Attacks | arXiv (co-authored) | 2025 |
| Project | Tech Stack | Impact | |
|---|---|---|---|
| ⚡ | JanSevaAI | React 19 · AWS Lambda · Bedrock · YOLOv8 · DynamoDB · S3 · CloudFront | Live AWS · 13 Lambda functions · 3 roles · SNS OTP |
| 🛡️ | LLM Prompt Injection Firewall | Python · ML Classification · Rule-based Filtering | 85% detection accuracy · real-time sanitization |
| 🔍 | GraphQL Anomaly Detection | LSTM · CNN · TensorFlow · OWASP tools | 92% accuracy · arXiv published |
| 🏗️ | Multimodal RAG — Construction Safety | PyTorch · SageMaker · AWS Lambda · S3 · Vector DB | Serverless · production SageMaker endpoints |
| 🤖 | AI Network Intrusion Detection | XGBoost · Autoencoders · LSTM · NSL-KDD · CICIDS2017 | Hybrid ML + DL · UNSW-NB15 benchmarked |
| Achievement | |
|---|---|
| 🧩 | 500+ LeetCode problems solved — strong DSA & problem-solving foundation |
| 📄 | Co-authored arXiv paper on AI-driven GraphQL anomaly detection |
| ☁️ | Live AWS deployment — JanSevaAI serving real civic use cases |
| 🔒 | 85% LLM attack detection — prompt injection firewall in production |
| 🏆 | Kaggle competitor — NVIDIA Nemotron Model Reasoning Challenge (LoRA fine-tuning) |


