Currently building voice AI at Qualcomm

I build AI systems,
models & products
that actually ship.

Hi, I'm Hemanth, a machine learning engineer with a real passion for speech recognition, natural language processing, and AI research that combines academic depth with practical, real-world applications. I joined Qualcomm in April 2026 as a Voice AI Systems Engineer.

My work involves designing efficient neural architectures, experimenting with models like Conformers and RNN-T, and building robust pipelines for multilingual and accented speech data. I've also contributed to open-source tools and enjoy turning research ideas into deployable systems that perform under real-world constraints.

I'm interested in low-latency voice interfaces, multilingual speech processing, and making human-computer interaction more natural. When I'm not coding or training models, I'm usually reading research papers or hitting the trails for a run.

See projectsExperience & CV
At a glance
Now
Voice AI Engineer · QualcommVoice AI · Hyderabad · since April 2026
Focus
Edge deploymentResearch & DevelopmentLarge Language Models
Education
B.Tech, AI & MLVellore Institute of Technology, Bhopal · 2020 – 2024
05 · Latest posts
02 Nov 2026
FlashAttention, Part 2: Softmax on the fly
One running maximum, one running sum, and the correction that keeps them in sync.
01 Nov 2026
FlashAttention, Part 1: Making softmax safe
Before we make attention fast, let's make sure its numbers behave: the attention shapes, why memory matters, and a numerically safe softmax.
12 Jul 2025
Knowledge Distillation
Mathematical derivation of distillation.