NBN Co is building an internal data science capability called ‘insightLab’ that is being put to work finding ways to improve its wholesale services.
The insightLab is understood to be only around six months old and is intended to act as a central resource to support NBN operations.
Its existence was revealed over Easter when the company unveiled plans to bring in “several data scientists to lead the research into advanced analytical insights and ML/AI-based solution opportunities".
“As the NBN is moving to the scale phase, in parallel, the telecommunications industry is undergoing a transformation driven by network-IT convergence, virtualisation, analytics and machine learning,” the network builder said.
“As NBN Co’s business accelerates, our challenges become increasingly complex whilst requiring responsive, informed decision-making and high levels of intelligent automation.
“Data scientists will be key to working through multivariate and ecosystem-wide problems, driving the thinking and communication of how NBN Co uses advanced analytics techniques to solve these problems in new ways.”
An NBN Co spokesperson would not be drawn on specific projects insightLab is being put to work on.
“NBN insightLabs has recently been established to incorporate data analytics to optimise our wholesale services,” the spokesperson said in a brief statement to iTnews.
NBN Co did indicate that the planned expansion of the resource may be linked to a desire to run more AI or ML services in production.
Recent research by iTnews has shown that enterprises are now starting to move beyond AI/ML proof-of-concepts as they seek to demonstrate that such solutions can scale and deliver higher returns.
However, the transition from PoC to production has been challenging for many early adopters, because the models have not been able to handle the sudden step change in scale.
In addition, many enterprises that are getting deeper into AI or ML projects face challenges in improving the accuracy of the models beyond around 80-90 percent.
NBN Co indicated it hoped insightLab data scientists will be able to write “sustainable, performant analytical code that drives repeatability of results and scalability of our analytics practices".