Welcome to My Portfolio

Professional Developer & Designer

View My Work

About Me

Hello! I'm Archan Kundu Chowdhury, a passionate computer science student with a strong interest in Machine Learning and Deep Learning, alongside hands-on experience in web development.

I enjoy building intelligent systems that combine data-driven decision making with clean, scalable web interfaces. My work focuses on applying ML and DL concepts to real-world problems while delivering user-friendly digital solutions through modern web technologies.

I’m constantly learning and experimenting with new frameworks, algorithms, and tools to stay updated with advancements in AI, data science, and full-stack development, aiming to bridge the gap between research and practical applications.

Enthusiastic about continuous learning, innovation and maintaining a disciplined lifestyle through running and regular workout

Skills

Machine Learning

NLP, Predictive Modeling, Classification, Sentiment Analysis, Supervised Learning, Unsupervised Learning, Reinforcement Learning

Deep Learning

Neural Networks (ANN,RNN,CNN), Transfer Learning, Transformers, Attention Mechanisms, Langchain, RAG

Programming Languages & Version Control

Python, Java, MySQL, HTML, CSS, JavaScript, Linux/Unix, Git/GitHub

Full Stack Development

React, Spring Boot, Flask, Tailwind CSS, Material UI, Streamlit, MySQL, MongoDb

Data Analytics

NumPy, Pandas, Matplotlib, Plotly, Seaborn, Data Visualization, Analysis

ML/DL Libraries

TensorFlow, PyTorch, scikit-learn, Hugging Face, Keras

Projects

Expression Classification
View on Kaggle

Facial Expression Recognition using EfficientNetB0

mplemented Facial Expression Recognition using Transfer Learning. A complete pre-processing and implementation is displayed along with a comparison of the simple CNN Model(56% accuracy) and EfficientNetB0(42% accuracy) and also Fine-Tuned EfficientNetB0, which got 66% accuracy on test data.

CNN Deep Learning Sequential EfficientNetB0 Fine-Tunning Transfer Learning Image Multi-Class Classification Expression recognition
AI Therapist
View on Kaggle

AI Therapist using Generative AI

React-Flask web application integrated with LLM through Langchain. Implemented RAG as vector database with security against LLM hallucination and Prompt Injection (under development).

React Flask Langchain RAG MongoDb Chatbot
Mental Health Chatbot
View on Kaggle

Mental Health Chatbot

Implemented using DialoGPT-small transformer with zero-shot sentiment analysis achieving 73% accuracy. Evaluated with BLEU & BERT scores for response quality assessment.

DialoGPT Transformers Sentiment Analysis NLP BERT-SCORE Evaluation
Fake News Classification
View on Kaggle

Fake-Real News Classification

Developed LSTM, Bidirectional LSTM & GRU models achieving 99.43%, 99.73%, and 99.71% accuracy respectively for binary classification of news articles.

LSTM BidirectionaLSTM GRU TensorFlow Binary Classification
Movie Recommendation System
View on GitHub

Movie Recommendation System

Complete NLP-based system suggesting movies using content-based filtering with intuitive Streamlit UI for seamless user experience.

NLP Content Filtering Streamlit Python
Spam SMS Detection
View on GitHub

Spam SMS Detection

Implemented using KNN, SVC, Naive Bayes, AdaBoost. Final Multinomial Naive Bayes model achieving 97% accuracy & 100% precision with Streamlit interface.

Machine Learning Naive Bayes Streamlit scikit-learn Classification
Online Bakery Shop
View on GitHub

Online Bakery Shop Website

Full-stack e-commerce application using SpringBoot backend, React frontend, and MySQL database with complete shopping functionality.

Spring Boot React MySQL Full Stack
Image Classification
View on Kaggle

Dog-Cat Image Classification using Transfer Learning

A complete CNN based Binary Classification. MobileNetv2 model is used for the classification with 97% test accuracy

CNN Deep Learning MobileNetv2 Transfer Learning Image Classification

Get In Touch