Ayush S Pangaonkar

AI/ML developer passionate about GenAI, LLMs, data science, and backend engineering. I build practical AI products, explore new technologies, and share what I learn along the way.

Mumbai , Maharashtra Joined on Oct 4, 2026 https://github.com/bluntjudg

Education

Tata Institute of Social Sciences (B.Sc, AI & ML), M.Sc Data Science at K J Somaiya (ongoing)

Pronouns

Ayush S Pangaonkar

Oct 11

Three Things I Found When I Revisited My Gender Classification Project

#machinelearning #python #beginners #datascience

2 min read

Want to connect with Ayush S Pangaonkar?

Create an account to connect with Ayush S Pangaonkar. You can also sign in below to proceed if you already have an account.

Create Account

Already have an account? Sign in

Ayush S Pangaonkar

Oct 10

My Near-Perfect R2 Was Fake: Debugging a Speed and Distance Model

#machinelearning #python #beginners #datascience

2 min read

Ayush S Pangaonkar

Oct 9

96.5% Accurate Spam Filter, and the 58 Spam Messages It Let Through

#machinelearning #python #beginners #datascience

2 min read

Ayush S Pangaonkar

Oct 8

Eight Emotions, 34,792 Tweets, and One Label I Don't Fully Trust

#machinelearning #python #beginners #datascience

2 min read

Ayush S Pangaonkar

Oct 7

When My House Price Model Hit 100% Accuracy, I Should Have Been Worried

#machinelearning #python #beginners #datascience

2 min read

Ayush S Pangaonkar

Oct 6

A Book Recommender That Nails Harry Potter and Fumbles Seven Plays

#machinelearning #python #beginners #datascience

2 min read

Ayush S Pangaonkar

Oct 5

Five Classifiers, One Dataset: What I Learned About Model Choice

#machinelearning #python #beginners #datascience

2 min read

Ayush S Pangaonkar

Oct 4

Subreach: Building Two AI Agents for Reddit

#ai #automation #javascript #llm

6 min read

DEV Community

Ayush S Pangaonkar

Badges

Writing Debut

Skills/Languages

Currently learning

Available for

Three Things I Found When I Revisited My Gender Classification Project

Want to connect with Ayush S Pangaonkar?

My Near-Perfect R2 Was Fake: Debugging a Speed and Distance Model

96.5% Accurate Spam Filter, and the 58 Spam Messages It Let Through

Eight Emotions, 34,792 Tweets, and One Label I Don't Fully Trust

When My House Price Model Hit 100% Accuracy, I Should Have Been Worried

A Book Recommender That Nails Harry Potter and Fumbles Seven Plays

Five Classifiers, One Dataset: What I Learned About Model Choice

Subreach: Building Two AI Agents for Reddit