Computer Science Engineering

Vishwakalyan
Patil

Machine Learning & Deep Learning Explorer | LLM & RAG Builder | Python Developer | Pre-Final Year CSE Undergrad

Currently working on LangChain Agents & RAG Systems, TFLite Edge Deployments, and first-authoring a research paper on Multi-Modal Deepfake Detection. Exploring Multi-Modal Learning, MLOps, and System Design.

Portfolio

Featured AI/ML Projects

Production-ready AI systems with measurable impact and scalable architecture

Multi-Modal DeepFake Detection

99.8% AUC • Multi-modal analysis • Research publication

Advanced deepfake detection system using EfficientNet ensembles, audio spectrograms, and frame-based classification. First-authored research paper under review with state-of-the-art performance.
Python
PyTorch
CNNs
Librosa
OpenCV
FastAPI
Kotlin
YouTube Q&A RAG Assistant

Real-time Q&A • Semantic search • Video transcript analysis

LangChain-powered RAG system enabling natural language queries on YouTube videos using extracted transcripts. Intelligent video content analysis with semantic search capabilities.
LangChain
OpenAI GPT
FAISS
Streamlit
YouTube API
Phishing Domain Detector

Real-time detection • REST API • Mobile integration

Real-time phishing URL detection system using lexical and host-based features. Deployed via REST API with Android integration for mobile security.
Python
Scikit-learn
BeautifulSoup
Flask
Kotlin
TFLite On-Device Image Classifier

On-device inference • Real-time prediction • Optimized performance

Custom image classifier optimized for Android deployment with quantization techniques. Achieves real-time prediction with minimal latency on mobile devices.
TensorFlow
TFLite
Android Studio
Kotlin
Quantization
AI Interview Simulator

LLM evaluation • Automated scoring • Interview preparation

Behavioral interview simulator using LLMs to evaluate and score candidate responses. Provides detailed feedback and improvement suggestions.
LangChain
Gradio
Transformers
NLP
Scoring
Firebase
FastAPI
Kotlin

Additional Work

Other Projects

Full-stack applications and mobile solutions with proven user adoption

Cricket Score Predictor

Match prediction • Statistical analysis • Sports analytics

ML model predicting target cricket scores using match context, regression models, and comprehensive team performance statistics.
Python
Pandas
Scikit-learn
Linear Regression
Quora Question Pair Similarity

Semantic similarity • Duplicate detection • NLP processing

NLP system classifying duplicate questions using semantic and syntactic similarity models with advanced text processing techniques.
Python
TF-IDF
XGBoost
Siamese LSTM
Faculty Research Portal

Institutional deployment • Research management • Live system

Institutional web portal for storing and displaying faculty achievements including papers, patents, and research work. Deployed for university use.
PHP
XAMPP
MySQL
HTML
CSS
JavaScript
Intelligent Transportation System

IEEE Conference • Smart city prototype • International presentation

Built a multilingual web platform for my Head of Department to promote road safety and smart transportation awareness. The website supported real-time translation in 14 languages, enabling inclusive access to international audiences.Presented at the IEEE Fellowship for the Intelligent Transportation Systems (ITS) WIE/YP Research Forum in Paris.
PHP
Firebase
MySQL
HTML
CSS
JavaScript

Academic Contribution

Research & Publications

Cutting-edge research in deepfake detection with state-of-the-art performance metrics

Advanced DeepFake Detection Using Multi-Modal Analysis
Under Review • Cybersecurity Journal • First Author

Developed a novel multi-modal approach combining facial landmark analysis, temporal inconsistency detection, and transformer-based architectures. The proposed method achieved state-of-the-art performance on benchmark datasets, significantly outperforming existing solutions while maintaining real-time inference capabilities.

0.998
AUC Score
99.2%
Accuracy
15ms
Inference
5M+
Samples

Recognition

Achievements & Awards

Recognition for technical excellence and leadership in competitive environments

KarNova S.G.Balekundri Institute of Technology

Awarded for developing CNNs and facial artifact analysis system with innovation and real-world applicability

Winner - Multi-modal Deepfake Detection System

Ingenious Angadi Institute of Technology

Secured 2nd place with improvements in video and image stream handling across platforms

Runner-up - Enhanced Deepfake Detection Project

Hack2Future - 24 Hour Hackathon

Built end-to-end misinformation detection tool combining Deepfake Detection with LLM-based Fake News Detector

Gogte Institute of Technology

IEEE Fellowship for the Intelligent Transportation Systems (ITS) WIE/YP Research Forum

Presented Intelligent Transportation System at global conference

International Presentation - Paris

Research Publication

Under review in cybersecurity journal with 99.8% AUC performance

First Author - Multi-Modal Deepfake Detection

Technical Expertise

Skills & Technologies

Proficient in modern AI/ML frameworks and full-stack development technologies

Python

C/C++

PyTorch

TensorFlow

Scikit-Learn

LangChain

TFLite

OpenCV

FastAPI

Flask

Hugging Face

Firebase

MySQL

GitHub

Get In Touch

Let's Connect

Open to discussing opportunities in AI/ML engineering, research collaborations, or innovative technology projects. Let's build something impactful together.

Email MeGitHubLinkedIn
Built with v0