Customer is worlds biggest telecommunication provider and customer wants to roll out the specific 5G customer who wants to churn from 4G to 5G. Predict customers who are going to churn and find out the factors that are leading to churn based on the customer feedback reports. Responsibilities:- Responsible for implementing data mining and statistical machine learning solutions to various business problems. Analyze large volumes of data to generate insights and actionable recommendations to drive business growth. Contributed to data mining architectures, modeling standards, reporting, and data analysis methodologies. Experience in Statistical Modeling, Time Series Analysis, Optimization, Data Mining, Machine learning techniques and algorithms, Cluster Analysis.

This was basically a Q&A system where user can ask questions and service will respond based on trained data. Technique applied - Used ChatterBotCorpusTrainer API to train the bot and google audio recognizer to sense audio question data & respond with correct answer.

Collected data about application/server issues from various telecom applications Analyzed the issue and respective solution and used them for application wise classification Also used them for various correlated issues and used to predict chain of issues Developed python scripts for monitoring servers and applications and collected data Used monitoring results data for predicting server/application issues

DAE (Data Analytical Engine) is a supervised multi-label text classification program briefly. In detail the main aim of DAE is to assign a set of pre- defined class to the unknown findings (Test Data) by training a set of classifiers efficiently on the already classified known data (Training Data). Here it is a multi-label classification problem and we are solving it by converting it to a multi-binary classification problem. A finding is a combination of a group of raw text columns and a group of categorical columns and each finding is identified by a unique finding_id and each finding can have multiple classes. Our aim is to use the training data for each class at a time, build the classifiers and classify test data. Responsibilities: Involved in Data extraction & dataset preparation. Conduct analyses and generate tables, listings, and Reports. Data cleaning and data transformation as per requirements

The OpenDaylight Project is a collaborative open source project hosted by The Linux Foundation. The goal of the project is to accelerate the adoption of software-defined networking (SDN) and create a solid foundation for Network Functions Virtualization. In Clustering we are trying to achieve High Availability, Performance and Horizontal Scaling. ODL supports 3 node cluster where it used Akka framework for electing its Leader and the rest nodes becomes Follower. Responsibilities: Evaluating Opendaylight Beryllium and finding clustering issues to improve the scalability functionality for 3 node to 5 node. Implementing tools for push 1million flows to Opendaylight controller and finding performance with 3 control Nodes and 5 Control node. Exploring the capacity of the controller when the switch count is increased from 1 to n, along with cluster scalability. Evaluating performance of the functionalities like resync and reconciliation functionalities. Fixing issues and contributing back to ODL. CoreJava, Shell scripts, Maven,Git

Customer is leading Analytics company mainly dealing with Image Detection, Ammunition detection and the software are deployed in schools and universities where we often here that there are mass shootouts in the US Schools/Colleges. Basic need is to stop those shootouts.

Customer is world’s biggest telecommunication provider and customer wants to roll out the specific 5G customer who wants to churn from 4G to 5G. Predict customers who are going to churn and find out the factors that are leading to churn based on the customer feedback reports

This was basically a Q&A system where user can ask questions and service will respond based on trained data.

Technique applied - Used ChatterBotCorpusTrainer API to train the bot and google audio recognizer to sense audio question data & respond with correct answer

Built machine learning model to predict positive & negative sentiment of tweets based on unigram word frequency features as well as other features like total words in documents, using Word2Vec model using genism library. Using Python script to download tweeter data. Pre-Processing was done using Python

CPS must maintain persistent custom variables at the session level and subscriber level that can be used in custom policies (Policies Tab in PB) and CRD table evaluation. Session level variables will exist for the duration of the specific session and subscriber level variables will exist until all sessions are closed.

Implementation of Alarms and Counters. Responsibilities: Ericsson ODL has some 16 counters and Alarms. Used JMX technology to implement the Alarms. Testing of the Alarms and counters. Contributing back to the ODL community in the Boron Release. CoreJava, Shell scripts, Maven,Git

To create Qemu VM and bring up a cluster Node where ODL will be running on these VMs using CBA. Responsibilities: Working on NFV using QEMU hypervisor to bring up the SDN cluster where OpendayLight Controller will be running in NB. Cluster includes Control Node and Payload Node which does HA, Clustering and Horizontal Scaling using Ericsson CBA. Static and Dynamic Installation of ODL in the PLs via Host Installation. Integration of Daily build to the Test Regression Suits run by the Regression Team. Worked on different Upgrade scenarios like Rolling Upgrade,SingleStep Upgrade and BFU using CBA framework. Troubleshooting issues if Upgrade is failing.

Worked in the RnD(SDN_CoE) Project of TechMahindra. We were developing Core Java based Applications for SDN . We are using Floodlight/OpenDayLight Controller along with some vendor specific library like Cisco OnePK to address the use cases. Have worked on the Bandwidth On Demand use case for the Cisco ONE Live held in June13 Responsibilities: We used the Floodlight as our controller and Cisco OnePK Java library for the application development. Floodlight Controller will take the decision on which path the routing should happen on a network based on highest bandwidth . In the abstraction side we used IOL and Cisco 819 router. Design and application development using OnePK to address the use cases like BoD. Integration of OnePK to an existing controller like floodlight. Worked on the following application development using OnePK: o OnePk Dynamic Topology Discovery o OnePk Routing Application o OnePk QoSPolicy Application

PMS (Performance Management Support) provides an interface to manage individual performance management scanners on WCDMA network elements (RNC, RBS and RXI). It collects statistics and recording files for active scanners on valid, connected nodes. It enables the concept of profiles. That is, managing a group of identical performance monitoring on nodes of the same type. Allows operator or external management system to set up and administer collection of performance management data from the WRAN and LTE networks. Schedules the collection and storage of performance management data from the nodes and issues notifications where performance data files have been collected. Responsibilities: Coding in Core Java, Review with senior member. Unit Testing using JUnit Implementation Support for any issues (FOA) Customer integration Customer Support Customer Product information Documentation