MACHINE LEARNING
-
Cleanse and normalize your data from various sources using our AI/ML: frameworks
-
Build models using your data and test them using pilot deployments
-
Carry out predictive modeling using your data and measure the outcomes
-
Predict outcomes and efficiently allocate resources and maximize ROI
Architecture & Assessment
Define the problem, solution roadmap and select the appropriate technology, tools and processes to achieve the objectives. Our services –
-
Problem and solution roadmap definition
-
Business use case identification and requirements definition
-
Data assessment
-
Technology Identification
-
Architecture Recommendation
Infrastructure Design
Implement the AI/ML data ETL pipeline and infrastructure for data collection and modeling using AWS, Azure or on-premise infrastructure deployment –
-
Data cleansing & profiling pipleline
-
ETL – Distributed data processing pipeline
-
Data archiving
-
Meta-data management
-
Integration with 3rd party data sources to user data pipelines
Development & Implementation
AI/ML models implementation –
-
Design & development of learning models
-
Algorithm development and benchmarking
-
Integration with existing enterprise data sources to integrate into models
Analytics & Visualisation
Leverage our expertise and experience working with various data visualization and analytics tools like QlikView, QlikSense, Azure ML and other 3rd party service providers. Our services include –
-
Statistical Analysis
-
Machine Learning
-
Data Modelling
-
Analytics
-
Data Visualization Solutions
AUTOMATED TESTING FRAMEWORK USING MACHINE LEARNING
Our client, a leading industrial devices manufacturer engaged us for development of GUI for touch screens on their industrial devices. The GUI development involved extensive integration and regression testing with hundreds of possible screen and event traverse paths across the GUI.
Challenges
-
The client’s IT team was budgeting and spending a lot of man-months for integration and regression testing of GUI screens
-
There was no way to ensure complete test coverage as new traverse scenarios kept opening up based on changes to the specifications and designs from engineering/product team
-
Creation of new test cases and execution of these test cases ended up pushing the release dates further delaying the projects
-
Inordinate amount of time was being spent on reviews of test cases
Need
Our client needed an automated test solution that could generate test cases based on revisions to screen designs and learn from defects discovered in the past test cycles. The framework required to have the capabilities to –
-
Generate new test cases
-
Integrate with automated test scripts to execute the test cases
-
Based on past defects data identify the test cases that needed to be included in test suite
Solution
Our team designed and developed an ML framework that was able to identify new test scenarios based on changes to screen design specifications. The framework was able to suggest a test suite based on past defects data from the integration and regression test execution cycles. Work is underway to enhance the framework using a score/weightage to address higher priority traverse scenarios.
Results
-
our ML engine improved the client’s test management process and reduced the manual inputs required in test case, suite design and execution
-
significant time-savings of up to 60% in with auto-generated test cases and test suite using our ML framework
-
a focused approach to integration and regression testing was achieved in response to the changes in screen design and specification
-
the ML framework utilized past defects data to identify potential breakage scenarios in screen workflows and ensured test coverage of these higher risk scenarios
Technology Stack
-
Scala, hadoop, tensorflow, Sagemaker, PyTorch, Presto, AWS RDS, Python