$10.0 per Hour
A performance driven data scientist with an aim to mine the hidden patterns from ocean of raw data. Ability to communicate and motivate team members to enhance strategic goals and bottom line objectives. Creative problem solving and troubleshooting skills complemented by meticulous attention to details that will result in the success of an organization by developing new applications or improving existing ones.
Tech Stack Expertise
- January 2017 - February 2023 - 6 Year
Alarm Dust Management In Telecommunication Network
- January 2018 - January 2019 - 13 Months
Build machine learning models like Apriori, FPGrowth in Python to identify main fault alarms from streaming fault alarms in 5G telecom networks that reduced ticket generation at Network operation centers by 50%.
Build ML pipelines to train the Machine learning models and validate the Machine Learning models using appropriate performance metrics like Support, Lift, confidence, and Leverage scores.
Promoted model transparency by making them interpretable and explainable for business needs and tracked experiments using MLFlow.
In this project we use Machine learning techniques to troubleshoot the fault alarms in telecommunication networks that can help the Network Engineers at Network Operation centers to easily identify root cause of fault and troubleshoot the fault alarms. Through this we intend to create end-to-end automation-based fault alarm management system in telecommunication network in real time.
Alarm Prediction In Telecommunication Networks
- January 2019 - January 2020 - 13 Months
Build Machine learning models like Random Forest, K-NN that can predict whether there is a fault in telecommunication networks based on the KPI given by sensors from the telecom network.
Build ML Pipelines to train the Machine learning models and validate the Machine Learning models using performance metrics like Precision.
Utilised hyper-parameter techniques like Grid search to improve the model performance by 10%
Enabled model transparency to business stakeholders and tracking of experiments using MLFlow
Project Description: In this project we use Machine learning techniques to predict the fault alarms in telecommunication networks that can help the Network Engineers at Network Operation centers to be able to prevent the network downtime due to these faults. Through this we intend to create end-to-end automation-based fault alarm prediction management system in telecommunication network in real time.
Customer Segmentation Of High Net-worth Individual
- January 2020 - November 2020 - 11 Months
Build clustering model to segment customer transaction behavior of different categories so that bank can offer better products for different categories of customers.
Evaluate the clustering models on basis of purity measures and collaborate with clients on the same.
In an investment bank there are lots of high networth individual customers for whom the bank advises on their wealth management. The investment banking company wanted to assess the spending behavior of these high networth individuals so that it can better offer its products (credit card etc) to different groups of customers.
AI Based Listener For Early Detection Of Heart Di
- January 2021 - May 2022 - 17 Months
Build machine learning models like LSTM, XGBoost that can detect the abnormalities in heart sounds in Paedatric children.
Evaluate the machine learning model on train and test sets and publish a research paper in IEEE
Project Description: Pediatric children sometimes will be born with abnormal heart conditions like a hole in the heart or their heart ventricles will be blocked. This causes a disturbance in the heart sounds which medically is referred to as arterial fibrations. Arterial fibrations are the main causes of heart attacks which results in the death of pediatric children. Here we build machine learning models that can detect these abnormal sounds of paedratic children thereby preventing deaths in paedratic children. The observations of the experiments are published as a research paper in IEEE conference.
Spam Detection For Emails Using Machine Learning
- January 2022 - April 2023 - 16 Months
1. Explored and built machine learning models like Logistic Regression that can detect whether the email is spam email or not.
2. Performed exploratory data analysis on the banks spam email data.
3. Evaluate the machine learning models using measures like accuracy, precision.
4. Provided support and maintenance of the ML models in production.
Project Duration: 1.9 Years
1. In an investment bank social engineering and spam emails results in loss for the company millions of dollars and reputation. In this project, we had to build spam classifier for a central European based customer that can detect and classify the emails as spam or not.
in Bachelor of EngineeringKarnataka University
- June 2011 - June 2015