Niruphan Rajendran


Niruphan Rajendran

Desire to present key achievements/learning externally and give back to the wider software development community at meet ups and conferences. Always strive to convert those success stories into mind share. Have implemented my innovative ideas and published them as papers,
– IoT Device Security Firmware Analysis
– Adopt NFT to Realize the Full Potential of Connected Devices
– Telemetry Performance Test in Connected World – IoT Smart Devices
– How to design your tool Infrastructure to test SAP products

Technical Expertise
SAP Cloud Migration Architect | Cloud Certified (Microsoft Azure – Azure for SAP Workloads Specialty (AZ-120), Azure Administrator Associate (AZ-104) | Smart IoT Devices | IoT Device Security | Telemetry Data Simulation/Ingestion | Cloud Migration Consultant | Intercepting Networks | Code Tuning | Code Profiling | Tool Orchestration | Mobile Apps Architect

Project Management & Consulting
Strategic Planning | Providing Solutions & Strategies – IoT, Cloud, Mobile, Data, ERP | Establishing key performance indicators (KPI) | NFR Elicitation | Team Development | Business Development | Demonstrating NFT services | Line Manager

Title:Unlocking the power of Machine Learning in Mobile NFT world


Machine learning (ML) is a branch of artificial intelligence (AI) which focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy. Itis an application of Artificial Intelligence which can automatically learn and improve from experience without human interference. Machine learning (ML) has transformed how we as humans interact with machines, technologies, and data. But it wasn't always so popular; what started out as a niche industry has grown into a billion-dollar market.Today, machine learning forms a part of every industry you could think of healthcare, finance, entertainment, retail, and manufacturing. It has become essential for businesses to adopt ML to boost revenue, cut costs, and automate operations.Here are some fascinating ML statistics to highlight its vital role in modern world.Netflix saved $1 billion due to its machine learning algorithm for the combined effect of personalization and content recommendations.Adding to this, 60% reduction in Google Translate errors was found when changed to GNMT—a translation algorithm powered by machine learning.The accuracy of Google's AI machine learning algorithm in predicting a patient's death is 95%.This also plays important role in Mobile Users of which 97% of mobile users are using AI-powered voice assistants.ML have been aggressively launched to serve Mobile application development.We all know that a mobile device has become an integral part of our life. Applications deployed on devices straggle a bit even, it’s a “Yucky” situation to us. Mobile NFT helps to a certain extent in overcoming these challenges. But the big question is, does it fix all the issues prevailing to each release in this DevOps world? The answer is “NO”. When it comes to Non-functional there must be more focus on data analytics which will be key to analyse reports/events successfully. Predicting the behaviour to understand the metrics in simpler and much easier way is a challenge. As we have huge chunks of data where we want to put together & provide a valuable outcome to business is a big ask. Below paper will focus moreon the smarter & intelligent approach to overcome all this challenge.ML based AINon-functionaltesting teams are facing an extraordinary challenge today is, post automated process, the overall release report must be consolidated for analysis and predicting the report by learning, training & evaluating it. In this DevOps world, there is a business need to share the predictive analysis to enhance the business process.From that perspective, the key asks are: 1.How would they ensure the coverage of the testing across Mobile OS versions in real world use case scenarios are performant? 2.How to alignthe business to focus on the required non-functionality of the entire Ecosystem under effective costTo overcome all these challenges, this paper will speak about the solutionfor each challenge from our point of view. In this Point of View[PoV], our focus is more on Data processing, Data Training and Data Evaluation. There are huge chunks of data residing with Non-Functional team. But for better evaluation & prediction we don’t have any specific technique or approach to take it up to Business. Adapting to the ML can help us fix those challenges & enhance the business. A smarter & intelligent approach to predict the user behaviour will help us understand metrics inmuch better way. Here we are proposing a newer way using TensorFlow framework by integratingML algorithms which will put together a huge chunk of data & provides a view on predicting the future behaviour. For example, consolidating response time across different bandwidths, devices,OS versions and predicting a future pattern of behaviour. This helps in analysing the flavour that needs to be tested, more importantly do we really need additional performance test and helps BA to align their businesses accordingly. By this way this reduces the overall cost that spent on different services.For successful implementation, itneeds to shift its focus at macro level and overcome below all the challenges:•Understanding the data•Data Visualization & Analysis•Data PredictionOur solution will focus on modern day Non-functional testing approach for Mobile NFT world using advanced ML technique, identify its key performance indicators and provide predictionfor end-to-end process. Solution Overview:1.Understanding the data:Collecting and organising data is a need of the hour. Key data will be collated and that helps in training and use it for prediction.2.Data Visualization & Analysis:Once data is loaded the next step is to train those data for better prediction. In this step we will visualize all those data to plot it in graph. 3.Data Prediction:The main factor of prediction comes over here. Here to predict right metrics for the Application with different parameters and combinations

After successful implementation of this solution,below listed benefits can be achieved•Algorithm for prediction will have an effective impact in delivery•This solution will save huge effort & cost for customers where they didn’t require to procure new devices for different testing•It helps Scrum Masters/Business to decide, do we really need additional performance test for release. This in turn reduces cost for customers.•Recommending approach will helpthe customer to predict the key non-functional information for different OS versions and will supportto tune App releases.Above solutions aided the business to derivekey non-functional metricsat different predictionsand this helps to visualise and decide performanceof the application. This diverse solution certainly benefits not only for the Mobile user’s performancebut also across several other type of users.