Large language model based QnA model for PCF system.
Graph based ML solution to observe and detect remote interference in 5G networks.
Model trees to predict the time to complete different milestones of network rollout project. Publication in COMSNETS 2023; Best paper award in MINDS workshop
Identify the mobile application and classifying the type of user actions (send/receive message, file transfer, voice or video calls) based on first few encrypted packets of TCP or UDP flow.
Inner source anomaly detection framework
Computer vision solution to detect the proper installation of network component in the tower for closure of 4G installatin. Object detection methods such as Yolo, SSD and Faster RCNN were evaluated to arrive at the final solution.
Active learning augument tagging tool for NLP ML tasking labelling.
LDA based topic modelling methods for open ended questions analysis from survey response.
Building ML based payment likelihood models for US Ford credit customers.
Classifying the customer interactions into appropriate CCCs and VFGs using word embeddings trained using word2vec technique.