Research Engineer, Speech and Language Lab, NTU Singapore Sep 2023 - Present

  • Under the guidance of Prof. Chng Eng Siong, my research subject is speaker diarization, specifically focusing on enhancing Target-Speaker Voice Activity Detection (TS-VAD) model.

  • Implemented a robust data pipeline for training the transformers based TS-VAD model on large wideband simulated data. Extended the model to handle arbitrary number of speakers, enhancing its versatility. Demonstrated the effectiveness of the TS-VAD model in achieving impressive results in the DIHARD3 challenge.

  • Proposed and integrated speaker counting, speech enhancement and speech separation techniques to solve far-field speaker diarization in noisy-condition.

Software Engineer, Special Clouds, Microsoft, Noida, India July 2021 - Aug 2023

  • Under the guidance of Prof. Chng Eng Siong, my research subject is speaker diarization, specifically focusing on enhancing Target-Speaker Voice Activity Detection (TS-VAD) model.

  • Implemented a robust data pipeline for training the transformers based TS-VAD model on large wideband simulated data. Extended the model to handle arbitrary number of speakers, enhancing its versatility. Demonstrated the effectiveness of the TS-VAD model in achieving impressive results in the DIHARD3 challenge.

  • Proposed and integrated speaker counting, speech enhancement and speech separation techniques to solve far-field speaker diarization in noisy-condition.

Software Engineer Intern, Outlook Calendaring, Microsoft, Bengaluru, India Jan 2021 - June 2021

  • Re‑engineered Outlook events and calendar APIs, shifting from a monolith to a micro‑services architecture using gRPC.

  • Designed and developed a traffic splitter tool to redirect traffic to the modernized APIs while also collecting metrics and telemetry for analysis.

  • Traffic splitter enabled safe and staged roll‑out and testing of modernized APIs worldwide while handling an average of 15 billion hits a month.

Software Engineer Intern, Evee Bot for Teams, Microsoft, Bengaluru, India May 2020 - July 2020

  • Evee intelligently manages calendar events and schedules work meetings inside Microsft Teams. My contributions empowered Evee to intelligently propose a meeting subject line whenever a real-time chat conversation indicates an intent to set up a meeting.

  • Trained, tested and fine-tuned Azure Language Understanding (LUIS) and keyword extraction models using the Enron Corpus. Proposed a bi-directional LSTM model for custom data.

  • Seamlessly incorporated the model with an exceptional BLEU score of 0.89 into the bot using Azure Bot Service.