░█████╗░██╗░░░██╗██╗░░░██╗░██████╗██╗░░██╗
██╔══██╗╚██╗░██╔╝██║░░░██║██╔════╝██║░░██║
███████║░╚████╔╝░██║░░░██║╚█████╗░███████║
██╔══██║░░╚██╔╝░░██║░░░██║░╚═══██╗██╔══██║
██║░░██║░░░██║░░░╚██████╔╝██████╔╝██║░░██║
╚═╝░░╚═╝░░░╚═╝░░░╚═════╝░╚═════╝░╚═╝░░╚═╝
ayush = {
"education": "B.Tech CSE @ KIIT University (2023–2027)",
"focus": ["Frontend Dev", "Machine Learning", "Cybersecurity"],
"languages": ["English", "Hindi", "Bengali", "Russian", "German"],
"currently": "Building cool stuff at the intersection of AI & Security",
"fun_fact": "I speak 5 human languages and 4 programming ones 🧠"
}Languages
Frontend
Backend & ML
Cybersecurity & Tools
|
FastAPI backend with fine-tuned BEN2 model for background removal and automated passport-style photo creation. Integrated Google Cloud Vision API for face detection & quality validation.
|
Logistic Regression + BERT transfer learning pipeline on IMDB dataset. Achieved 89% accuracy and improved F1-score by 12% with fine-tuned BERT. Deployed as REST API.
|
|
Time-series forecasting with LSTM on 5 years of historical data. Feature engineering with RSI, MACD, Bollinger Bands improved accuracy ~18%. Deployed via Streamlit dashboard.
|
ML pipeline (Random Forest, XGBoost, LR) for credit default prediction. Applied SMOTE for class imbalance, achieving AUC-ROC of 0.91. Inspired by JP Morgan Quantitative Research methods.
|
- 🥇 Winner – CPC HackerRank Programming Contest
- ☁️ Google Cloud Skills Boost – Generative AI, Responsible AI & Data Analytics
- 🧠 Deep Learning Specialization – DeepLearning.AI (Intro to Deep Learning)
- 📊 Quantitative Research Job Simulation – JP Morgan Chase & Co. (Forage)
- 🔐 Cybersecurity Intern – Wayspire (Apr – Jun 2024)
"The quieter you become, the more you can hear — in code and in life."