Our dataset contains the sample of candidates of a company who is facing challenge that about 30% of the candidate who accept the job offers but do not join the company. So with the help of Decision Tree algorithm, ‘gini’ as criterion, I build a model to predict the likelihood of candidate joining the company.

Objective: A health condition when liquid is accumulated in lungs, with the help of neural network based model we tried to classify x-ray image of chest whether is person have pleural effusion or not. 

Solution: Training CNN+LSTM, CNN+GRU, TL+LSTM, TL+GRU, Conv3d, OCR.Methodology: Defining different callbacks like EarlyStopping, ReduceLRonplateau, Model checkpoint, adam is chosen as optimizer.Developing different model architecture for training.Key Achievement: Model evaluation with the help of validation accuracy and validation loss achieved 81% accuracy on validation data

Solution: Training BERT, Custom Spacy, SentenceTransformer, Word Embedding, BiLSTM.Methodology: Spacy and BERT models for entity extraction, identification of similar tags and issues using SentenceTransformer, custom modifications on the BERT, word embedding, intent recognition.Key Achievement: Users are being provided with relevant hashtags suggestions which aim to enhance work efficiency and assist in categorizing issues effectively.

Solution: Using exploratory data analysis and machine learning, We need to find possible feature which effect lead conversion.Methodology: Pre-processing data and feature engineering with the help of pandas & numpy. EDA with the help of matplotlib and seaborn. Build Logistic Regression model with the help of scikit- learn.Key Achievement: Derived top 10 feature which effect lead conversion rate and build Logistic regression model with 91 % sensitivity

Solution: Training resnet18 architecture on x-ray image with user-defined callbacks.Methodology: Pre-processing image data and data augmentation with the help of keras. Performing Ablation experiment in small chunk of data to evaluate resnet18 architectureKey Achievement: Developed a model with AUC score 87 % on validation data.