Deepika Gandla
I'm a final-year B.Tech student in Computer Science and Engineering, specializing in Data Science. I build complete AI applications from training models to deploying working systems that people actually use. I work across the full stack: train models with PyTorch, build Flask APIs, create React frontends, and deploy everything so it runs smoothly. I've shipped a GenAI travel planner, a fake news detector with 98% accuracy, and several web apps handling real user traffic. Right now, I'm experimenting with AI agents and automation, building systems that don't just respond, but actually solve problems on their own.
About Me
A bit about me
I'm in my final year of Computer Science and Engineering (Data Science) at Sreyas Institute of Engineering and Technology. I got into AI because I was curious, everyone was using these AI tools, and I wanted to know how they actually work.
So I started digging. Reading, watching videos, experimenting with different tools. But I realized pretty quickly that understanding AI wasn't enough. Models sitting in notebooks don't help anyone. They need to become actual applications people can use.
That's why I focus on full-stack AI development. I work across the entire pipeline—training models, building APIs, creating frontends, and deploying systems that actually run in production. I don't just want to build demos, I want to build things that work.
The best part for me? Hitting deploy and watching something go live. Taking an idea from research to a working product that's what gets me excited.
Experience
Gen AI Intern
SURE ProED Jan 2026 - Present- Built an AI chatbot using RAG (Retrieval-Augmented Generation) to pull accurate answers from private documents, making it much easier and faster to find specific information.
- Created a Semantic Search system using Gemini that understands the actual meaning of a question instead of just matching keywords, helping users get more relevant results instantly.
- Currently developing Agentic AI and LLM workflows using Gemini 2.5 Pro that can reason through complex tasks on their own instead of just giving simple chat responses.
- Focusing on making these AI tools stable and reliable, ensuring they provide correct, real-time answers with zero lag so they are ready for real-world use.
Data Science with GenAI Intern
Innomatics Research Labs Nov 2025 - Present- Worked with Python and Advanced Statistics to analyze large datasets and find hidden patterns, cutting data preparation time by 20%.
- Handled the end-to-end process of moving AI models from research notebooks to live web applications using Flask, making them ready for real-world tasks.
- Currently building GenAI and Agentic AI systems that can independently think through and solve multi-step problems.
- Applying MLOps practices to manage the AI model from start to finish, ensuring the tools stay reliable and run with zero lag in production.
Skills
Technologies I work with
Programming Languages
Web Development
AI & Machine Learning
Tools & Practices
Projects
Building things that actually help
from sediffgen import TopicModel
import torch
model = TopicModel(
n_topics=50,
embedding_dim=768
)
model.fit(corpus) # Semantic diffusionSEDiff-Gen: Diffusion-Guided Topic Modeling Framework
Developed SEDiff-Gen, an AI framework that automates theme discovery in large text collections. By combining SBERT embeddings with Semantic Diffusion and HDBSCAN clustering, the system outperformed traditional LDA models with 30% faster processing and a Topic Coherence score of 0.58 (Cv).
Built using PyTorch, it features a custom evaluation suite and a visual dashboard that allows researchers to explore and visualize hidden themes across complex datasets instantly.
const itinerary = await
ai.generate({
model: "Gemini-1.5-Pro",
tools: ["weather",
"translate"]
});Travel Itinerary Generator Using AI
Built a full-stack AI travel assistant deployed on PythonAnywhere that generates personalized trip plans in under 10 seconds. Using the Gemini 1.5 Flash API, the platform cuts manual vacation planning time by around 70%. Pulls real-time weather updates via the Visual Crossing API and supports multiple languages through the Deep Translator library.
Developed with Flask and SQLAlchemy, featuring secure user authentication and a saved itinerary database.
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import PassiveAggressiveClassifier
vectorizer = TfidfVectorizer(max_features=5000)
model = PassiveAggressiveClassifier()
accuracy = 0.985 # 98.5% accuracyFake News Detection
Created a fake news detection system trained on 44,898 news articles that hits a 98.5% accuracy rate. It uses TF-IDF Vectorization and a Passive Aggressive Classifier to perform fast, real-time text analysis on news headlines.
Built with Scikit-learn, the tool flags misleading content immediately and makes it easy for people to check if news is real.
const timer = setInterval(() => {
if (timeLeft-- === 0)
nextQuestion();
}, 1000);Quiz Application With Timer
Designed a responsive quiz tool hosted on GitHub Pages that runs 25 questions per session with zero lag.
Built with HTML5, CSS3, and JavaScript, it handles real-time scoring, countdown timers, and instant color-coded feedback. The tool stays smooth on all devices and shows a clear performance summary at the end.
Certifications
What I've learned and certified in
STTP on Java Full Stack using React.js and AI
Brainovision Solutions India Pvt. Ltd
View Certificate
Let's Connect
Let's chat about projects, collaborations and opportunities!
Start a Conversation
Actively looking for Full Stack AI Developer or Data Science roles where I can build things that actually matter. If you have an interesting problem or project, I'd love to hear about it.