Research Engineer
I build speech & audio ML. I ship on-device AI for TVs. I design agentic workflows.
I'm a research engineer at LG Soft India in Bengaluru. Most of my work sits where speech and audio machine learning meets the limits of consumer electronics: models that have to run on a TV, in real time, on noisy broadcast audio.
Recently I designed a multi-agent system that migrated about 4,000 GStreamer patches, with only about 46 needing a human to step in. I also lead a task force bringing AI and agentic tools to a 500-engineer department. Alongside product work I built BLEEP, a word-level profanity speech dataset accepted at IEEE SLT 2026, to be presented in December.
Before LG I built full-stack products at Trustgrid.AI, co-founded an education startup, and studied electrical and computer engineering at Georgia Tech.
LG Soft India
Feb 2023 – now · Bengaluru
Trustgrid.AI
Feb 2022 – Feb 2023 · Bengaluru
Vantir
Dec 2019 – Feb 2022 · Bengaluru
Intel
Jan – Jul 2018 · BengaluruProjects at LG Soft India, by area
Choose a project to see the details.
LG Soft India
Port LG's ~4,000 downstream GStreamer patches from 1.24 to 1.28, a job that usually takes about ten person-months.
Speech recognisers trained on clean read speech under-recognise profanity, which makes live broadcast filtering unreliable. BLEEP is a word-level English dataset with phonetically confusable "hard negatives", recorded in neutral and expressive registers, with a benchmark across SSL and Whisper models.
Illustration: one profane word located and muted inside a speech clip.