Rahul Bhagat

Experienced data scientist with an 8-year proven track record in extracting valuable insights from complex datasets. Proficient in leveraging advanced Generative AI, statistical methods, and machine learning algorithms to drive data-driven decision-making. Adept at communicating technical concepts to non-technical stakeholders and collaborating with cross-functional teams to deliver impactful solutions.

• Date of Birth: 08/31/91
• Gender: Male
• Marital Status: Single/unmarried

E-Surveillance Portal-Computer Vision Image Classification Worked on Computer vision project to classify images as Cleaned and Uncleaned(Containing garbage) at Atms sites.Used Mobilenet model to accurately classify images with accuracy 96%.Used available images to train model.Deployed the model in Production server.Solved issues related to it on production server.Retrained model for data shift and redeployed it using CICD pipeline.Used Jupyter,Azure machine learning studio for training purpose.

Cash Dispence Forecasting

Worked on Cash dispense forecasting usecase. Worked on Timeseries data of ATMs to find out cash dispense forecasting for the entire month. Used Statistical technique to filled the missing and erroneous data using business logic. Tried diifferent timeseries and Statistical models to get better forecasting. Got an accuracy of 94% using Sarimax model on forecasting and actual data.

ATMs Fault Forecasting

Worked on ATMs fault detection project. Gathered data ,understand the data using business logic. Processesed timeseries data for missing values. Used SMOTE technique to fill imbalanced data. Tried different timeseries and statistical technique for better accuracy. Used LSTM model to train the model. Got an accuracy of 92% for Errors/fault forecasting with actual data. Presented a obtained results with the client.

Sentiment Analysis

Worked on the Reviews dataset to analyse the text sentiment and find out the sentiment of each statement. Used textblob ,nltk,library to get polarity of texts and deployed various algorithm like DL,adaboost,xgboost,and svm to get the maximum accuracy. Used ensambling method to get the model stability. This helps client to make judgement to drive their new ongoing market branding.

Subjective answer Evaluation

Developed Subjective answer Evaluation model based on Nlp and LLM architecture using Openai api.

Sales Forecast-Time series Analysis

Worked on data set to find out 6 month prediction for business. Performed feature scaling, used machine learning models, metric in order to compare model and choose the one with optimum result (Python & Machine Learning).

EDG Level IV Prediction

Worked on EDG level IV Prediction Project of McD. Training datasets include 15-16 Market data of different countries. in Databrics environment Worked on data analysis and Feature engineering,Data augmentation techniques to curate data to utilize for the model training purpose. Tried different alogorithms for better accuracy and generalized model. Finalized SVM model for best accuracy and generalization. Optimized model for 93% accuracy. Produced prediction results for Unseen 18 new markets. Deployed model in production environment and Comitted code to Github.

I love to visit natural places and watch Sci-fi movies