Open to AI/ML roles · mid-2026

Hi, I'm Ganesh Galbale.

I build AI systems that see, reason, and explain themselves.

See my workSay hi

The person behind the models

Final-year CS student at Bennett University (CGPA 8.62/10). I care about building AI that's interpretable, not just accurate.

My deepfake detector runs ResNet-34 with Grad-CAM explainability. My CLIP-based fake-news model is IEEE-published. Currently at Cog Culture doing gen-AI in production.

99.21%

Deepfake Detection Accuracy

✓

IEEE Published

8.62 / 10

CGPA

Education

Bennett University

Bachelor of Technology in Computer Science

· CGPA: 8.62 / 10

Machine Learning, Deep Learning, NLP, Computer Vision, Optimization Algorithms, Business & Data Analytics

Lal Bahadur Shastri Sr. Sec School

Higher Secondary (CBSE)

· 83%

Experience

Cog Culture

AI Intern

Building generative AI pipelines and integrating LLM-based tools into production workflows. Real deadlines, real users.

Advaita, Bennett University

Social Media Head

Led creative strategy for the university music club. Ran campaigns, coordinated a team, executed live events.

IEEE Publication

Published in IEEE. 90.90% accuracy. One of two co-authors.

View paper →

What I work with

No fluff — things I've used to ship real results.

Languages

What I write in daily

PythonSQLC++

Frameworks & Libraries

The ecosystem I actually use

PyTorchTensorFlowKerasScikit-learnStreamlitNext.js

Domains

Where I've put in real hours

Machine LearningComputer VisionDeep Learning (CNNs, RNNs, Transformers)Natural Language ProcessingGenerative AIData Analysis

Tools

Dev environment staples

GitDockerJupyter

Things I've actually built

Not toy demos — models with accuracy numbers and papers.

Deepfake Detection with Explainable AI

IEEE Published

Trained ResNet-34 to spot deepfakes at 99.21% accuracy — then built Grad-CAM, LIME, and Integrated Gradients overlays so you can see exactly which pixels the model is looking at. Accuracy without explainability isn't enough.

PyTorchOpenCVResNet-34CaptumLIME

Multimodal Fake News Detection

Fake news spreads through memes as much as through text. I fused CLIP image-text embeddings to catch both modalities at once — hitting 91.2% on the Fakeddit dataset.

PythonPyTorchCLIPScikit-learn

AI-Powered Fact-Checking App

A fact-checker that actually reads your PDFs and cross-checks claims against live web data using Gemini + Tavily. Catches both the subtle rewrites and outright fabrications.

PythonStreamlitGemini APITavily API

Say hi.

I'm actively looking for AI/ML roles starting mid-2026. If something you're working on sounds interesting, let's talk.

galbaleganesh@gmail.comlinkedin.com/in/ganeshgalbalegithub.com/GaneshGalbale

© 2026 Ganesh Galbale. Made by Ganesh, not entirely by AI 😄