Home

My Projects

RUL Inference Engine

Microservice-based Remaining Useful Life (RUL) prediction system using LSTM trained on NASA C-MAPSS data. FastAPI powers the inference backend; Streamlit provides a user-friendly UI.

FastAPI Streamlit LSTM Python
#predictive-maintenance #remaining-Useful-life
🔗 See Project

Predict Calories

A microservice-based application that predicts calories burned during exercise using user input. Built with Streamlit (frontend) and FastAPI (backend), and powered by a machine learning model trained on the Kaggle Playground Series S5E5 dataset.

FastAPI Streamlit XGBoost Python
#caloreis-burnt-prediction #deployment
🔗 See Project

ModernBERT for Natural Language Inference (NLI)

Fine-tuned version of answerdotai/ModernBERT-base on 30k samples from the SetFit/mnli dataset for Natural Language Inference. Classifies relationships between sentence pairs: entailment, contradiction, or neutrality. Built using Hugging Face Transformers and PyTorch.

ModernBERT Hugging Face Transformers PyTorch Token Classification Python
#nli #transformers #modernbert
🔗 See Project

BERT for Named Entity Recognition (NER)

Fine-tuned BERT-base model on the CoNLL-2003 dataset for Named Entity Recognition (NER). Achieves over 94% F1 score in identifying entities such as persons, organizations, and locations. Utilizes Hugging Face Transformers with token classification head.

BERT Hugging Face Transformers Token Classification Python
#nlp #named-entity-recognition #transformers
🔗 See Project

Amazon Product Listing Optimizer

Develped an agentic AI application for optimizing Amazon product listings using language models and dynamic tool invocation. Built with LangChain for agent orchestration and deployed via Streamlit.

LangChain OpenAI API Streamlit
#agentic-ai #ecommerce
🔗 See Project