This Project involves analusing the data being created by several network devices of various vendors of IBM. Data includes recording the node availability and different events happened during the run. My role in the project requires conversion of the existing python scripts to pyspark scripts. This is a challenging task as i had to understand several lines of code developed by another developer,understand the logic and replicate in pyspark.

various tasks involved in the project are:

1. Data pull from source Cassandra database

2. Data processing using pyspark (Analytics engine)

3. Data load to target Cassandra database.

This project is all about identifying the route cause analysis for ticket to be raised using different machine learning/Artificial intelligencetechniques. My responsibilities include

• Design the way to approach the problem, brainstorming with the team.
• Perform the ground work required to develop the idea to present to the client by developing some knowledge graphs
• Assist developers in case of dead locks.

It helps the Business to understand the areas of bottle necks in different applications. and involves 2 steps of implementation –

Step1: Developing a probability matrix to give an overview of confidence levels of different child servers that are responsible for a given issue.

Step2: Developing a semantic domain knowledge library to bring out the relation between different issues of parent and child servers.

This project requires analysis of server logs for identifying the route cause of outages. My role involved following responsibilities

• Conversion of log data of a server into a data frame
• Identify driving factors that are responsible for outages during the server run time.
• Develop data models using Multiple Linear regression, Keras
• Provide insights on root cause identified in the analysis.

This project involves had 2 applications

Part1: Text Analytics - KT Advisor

usage of oozie workflow to invoke text analytics model of clustering including mysql database and various big data technology tools like sqoop, Solr, Apache Tika, Spark. My role was a developer and involved in below activities:

• Conversion of python flask APIs to REST APIs
• Modifying existing workflow of oozie to minimize the overall workflow time
• Performing enhancements to fix issues in application

This app helped IT managers understand frequently occurring issues and identify the areas where maximum time is being spent in resolving issues, so that high level preventive measures are initiated.

Part2: Mobile Dashboard Analysis

This application involves automating complete data transfer from staging tables to Mobile dashboard using python. which includes fetching data from staging tables, converting the data to json format, committing the data to Git and triggering emails with appropriate message about job completion/failure to entire team.

This has almost reduced 100 hours of manual work per month for the team and benefitted the team to focus on other areas of development.

This project works on developing Classification models for classifying different classes of incident data (text data) using various machine learning modelling techniques like Random Forest, K-NN, Latent Semantic (LSA) Analysis. It involves usage of supervised machine learning models.

I am a Data analyst/Data modeler/developer wherein i have developed various models to classify the data into different classes and also fine tuned the models for better accuracy by identifying key features responsible and compared the accuracy with IBM WATSON Natural language classifier API.

It is all about identification of hidden patterns in metrics data of servers. My role involves in brainstorming with the team about the approach to followand development Python scripts to observe server metrics over a period and identify the unusual behaviour. This requires good knowledge of the data behaviour and proper variable selection for analysis. Thus, forward step wise regression is picked for variable selection.

Obtained results are then exported to Tableau to visualize. Using the patterns Identified, upcoming unavailability of the server is informed to the team to take preventive measures.

Automated complete data transfer from staging tables to Mobile dashboard using python. This includes fetching data from staging tables, converting the data to format, committing the data to Git and triggering emails with appropriate message about job completion/failure to entire team. This has almost reduced 100 hours of manual work per month for the team and benefitted the team to focus on other areas of development.

It is an account which supports multiple clients under one roof. Project is all about Extracting from source Cassandra DB, Processing using a customized analytics engine built on pyspark and loading the data into target Cassandra DB. My responsibility was to convert all the existing python scripts to pyspark scripts.

Anomaly detection Identification of hidden patterns when a server is being unavailable for operation. Developed Python scripts to observe server metrics over a period and identify the unusual behaviour. This requires good knowledge of the data behaviour and proper variable selection. Thus, forward step wise regression is followed during variable selection. Obtained results are then exported to Tableau to visualize. Using the patterns Identified, upcoming unavailability of the server is informed to the team to take preventive measures.