I'm a Computer Science student focused on building AI-powered, production-ready systems. I design machine learning pipelines for prediction and NLP tasks, build full-stack applications with React and FastAPI/Flask, and integrate real-world APIs to ship products end-to-end — from AI travel planners to disaster early-warning platforms.

name       : Anshu Gupta
role       : AI/ML Engineer · Full Stack Developer
college    : Graphic Era Deemed to be University (GEU), Dehradun
cgpa       : 9.10 / 10
current    : DII Intern @ National Informatics Centre (NIC)
focus      : Machine Learning · NLP · React + FastAPI/Flask · REST APIs
interests  : AI-Powered Products · Predictive Systems · Real-time Applications

Languages

Frontend & Backend

AI / ML & Data Science

Databases

DevOps & Tools

An AI travel planner that generates personalized itineraries with route optimization and budget estimation.

Built during my internship at NIC, Smart Travel Planner uses Llama 3.3 70B to generate personalized trip plans, ranking attractions and optimizing routes based on user preferences, time, and budget. The backend integrates live data from Google Maps, Places, OpenWeatherMap, and Groq APIs to keep recommendations grounded in real-world conditions.

Core Modules:

Architecture:

Frontend  →  React.js
Backend   →  FastAPI · SQLite
AI Layer  →  Llama 3.3 70B (Groq API) · Gemini
Data      →  Google Maps API · Places API · OpenWeatherMap API

An AI-powered disaster management system with real-time monitoring and forecasting.

Built under faculty supervision at GEU, DisasterSphere trains a Random Forest model on 50K+ records to predict disaster risk at 85% accuracy, visualized through live heatmap overlays. The platform features role-based access, JWT-secured REST APIs, and processed 2,000+ daily requests during testing.

Core Modules:

Architecture:

Frontend  →  React · Leaflet.js
Backend   →  Flask · SQLAlchemy ORM · JWT
ML        →  Scikit-learn (Random Forest)
Data      →  Firebase · Gemini API

A desktop phishing-detection application benchmarking seven ML and DL models.

PhishShield is a Tkinter desktop app that evaluates seven machine learning and deep learning models for phishing URL detection, with a full preprocessing and ensemble-learning pipeline. Model performance is compared across Accuracy, Precision, Recall, and F1-score to select the strongest classifier.

Core Modules:

Architecture:

UI        →  Tkinter
ML        →  Scikit-learn · XGBoost · TensorFlow/Keras
Evaluation → Accuracy · Precision · Recall · F1-score

An intelligent path-optimization system for domestic aviation route planning in India.

Designed to improve on traditional flight-path planning methods, this C++ system models optimized routing logic for domestic aviation networks.

• 🎓 CGPA 9.10/10 — B.Tech Computer Science & Engineering @ Graphic Era Deemed to be University
• 💼 DII Intern — National Informatics Centre (NIC) | Shipped an AI-powered itinerary platform integrating multiple live APIs
• 🔬 Summer Research Intern — GEU | Built an AI disaster management system handling 2,000+ daily API requests
• ☁️ AWS Certified — Machine Learning · Cloud Computing
• 📈 Google Data Analytics Professional Certificate · Data Science Certification
• 🏅 Hackathon Participant — Google Solution Challenge, FOSSHack, TechSprint-48, OpenHack 2025, Code Carnage, Graphethon

const currentFocus = {
    learning:   ["Advanced NLP", "LLM-powered Applications", "System Design"],
    building:   ["AI-driven Products", "Full-Stack ML Systems", "Real-time APIs"],
    exploring:  ["Agentic AI Workflows", "Cloud Deployment", "Model Optimization"],
    improving:  ["DSA Problem Solving", "Model Evaluation Rigor", "Production ML Pipelines"]
};

Built with precision · Driven by curiosity · From anshugupta06