Apr. 2021 - Oct. 2022Customer churn prediction for insurance:-
The client wants an understanding of which of his customers is most likely to churn.
Here we analyzed data to detect customers who are likely to cancel their premiums as the business model is based on premium. we are evaluating through a machine learning model that customers are most at-risk of churning so the client can use some preventive measures to not let that happen.
As the whole project contains sklearn pipeline model to save the model and the preprocessing in library from which is deployed in AWS by use of sagemaker the inference is accessible by AWS lambda and UX/UI design by the Platform team The technology used:-
- language:- - python
- algorithm:- - Gradient boosting
- Deployed using AWS Sagemaker - It also includes containerization and endpoints.
- Getting output using AWS Lambda and via UX/UI consol
ONE BCG - ONE Business Consulting GroupMachine Learning Engineer
Oct. 2018 - Apr. 2021Chandigarh Area, IndiaManaging Data Pipeline
Data Analytics & Reporting
Face Recognition Using Computer Vision Attendance System Using Facial Recognization, through Image Data trained into the Model provided for each employee.
In-Logs and Out-Logs per Employees in the office maintained using the same.
Technology Stack Used:
Python
Tensorflow-GPU
Cuda
Pickle
Resnet v1
Transfer learning
SSDN mobilenet
MTCNN
OpenCV PROJECT REPORTS VIA EMAIL USING ETL
Problem Statement based Data Extraction through MySQL using Python.
Storing & Transformation of the retrieved Data to HTML Format.
Data Packaging into Email and sending to the designated Person through Amazon SES. Advance Care Number Plate Recognition, the problem is to open the gate in real-time so that Car / Bike need not stop or wait at the Entrance. It is basically for the Employee associated with the Organization. Moreover, it will help the Security Personnel to keep a record of the Parking so that it won't impact the parking of VIP.
Python
Tensorflow-GPU
Cuda
Pickle
Transfer learning
Keras
OpenCV
YOLO V5 custom deep text OCR Customer churn prediction:-
Client wants an understanding of which of his customers is most likely to churn.
Here we analyzed data to detect customers who are likely to cancel their subscriptions as the business model is based on subscription. we are evaluating through machine learning model that customers most at-risk of churning so the client can use some preventive measures to not let that happen. Technology used :-
- language :- - python
- algorithm :- - Random forest. - XGBoost - lime to reach the granular level.
-Data extraction:- - Worked in postgresql for data extraction.
ConstalyticsData Science Trainee
Feb. 2018 - May. 2018Mohali punjabWorking on image classifier and also working on object identification in a video Comprehensive understanding of deep learning
Proficient in Python CREDIT CARD FRAUD DETECTION
Predicting whether a person could be a potential Defaulter or not, based on previous Dataset of 6 Months Credit Card Billing.
Dataset was cleaned, normalized, with proper Analysis to create a better prediction model. NEURAL NETWORK BASED IMAGE CLASSIFICATION
Image Classification upon a set of Images to Detect if the Image is that of a Cat or a Dog. Technology Stack Used:
Python
Matplotlib - Visualization
Numpy/Pandas - Data Management
Tensorflow - Understanding of Neural Networks