Profile banner
Akshit Arora

Akshit Arora

Senior Solutions Architect @ NVIDIA | Scalable Generative AI Systems

Daly City, California
San Francisco Bay Area
Joined January 2025

Network

3.8K connections
🌟
NVIDIA AI Leadership
💻
NVIDIA AI Engineers
📈
NVIDIA Product GTM
🏗️
NVIDIA AI Architects
🧠
AI ML Researchers
🏔️
...
🧑‍🏫
AI ML Educators

Summary

Akshit is a leading expert in scalable Generative AI systems at NVIDIA, focusing on large language models (LLMs) and text-to-speech (TTS) applications. He actively contributes to the development and deployment of advanced conversational AI models, particularly those designed for multi-speaker and multi-lingual capabilities, and shares his knowledge through publications and technical blogs on platforms like NVIDIA Developer and AWS. github+3
With a strong academic background from the University of Colorado Boulder, Akshit's research spans cognitive science, climate prediction, and AI for education, demonstrating a diverse application of deep learning to solve complex real-world problems. His Google Scholar profile showcases his contributions to educational data mining and interactive simulation tools. github+3
Akshit is a dedicated mentor, actively guiding students and professionals on career paths in deep learning, machine learning, and navigating the tech industry. He participates in mentorship programs, gives talks at universities like his alma mater, and shares insights through podcasts and webinars. github+1

Work

Education

Writing

Running NVIDIA NeMo 2.0 Framework on Amazon SageMaker HyperPod

March 1, 2025

AWS Machine Learning Blog post on deploying the NVIDIA NeMo 2.0 Framework on Amazon SageMaker HyperPod.

Favicon imageaws.amazon.com

Accelerate your generative AI distributed training workloads with the NVIDIA NeMo Framework on Amazon EKS

July 1, 2024

AWS Machine Learning Blog post on accelerating generative AI distributed training workloads using the NVIDIA NeMo Framework on Amazon EKS.

Favicon imageaws.amazon.com

Large scale training with NeMo Megatron on AWS ParallelCluster using P5 instances

May 1, 2024

AWS HPC Blog post detailing large-scale training with NeMo Megatron on AWS ParallelCluster using P5 instances.

Favicon imageaws.amazon.com

NVIDIA Speech and Translation AI Models Set Records for Speed and Accuracy

March 1, 2024

NVIDIA Tech Blog post on NVIDIA's speech and translation AI models achieving record-breaking speed and accuracy.

Favicon imagedeveloper.nvidia.com

The Curious Perspective

February 1, 2024

Personal blog launched on Substack, where Akshit captures raw thoughts around photography, research, and life.

Favicon imagearoraakshit.substack.com

Scaling NVIDIA’s Multi-speaker Multi-lingual TTS Systems with Zero-Shot TTS to Indic Languages

January 1, 2024

Research on scaling NVIDIA's Multi-speaker Multi-lingual Text-to-Speech systems using zero-shot TTS to Indic Languages. Presented at IEEE ICASSP 2024. This work won 1st place on ICASSP’s Signal Processing Grand Challenge (LIMMITS 2024) Track 3.

Favicon imagearoraakshit.github.io

VANI: Very-lightweight Accent-controllable TTS for Native and Non-native speakers with Identity Preservation

March 1, 2023

Research on a very-lightweight accent-controllable TTS for native and non-native speakers with identity preservation. Presented at ICASSP 2023. This project stood 1st in terms of speaker identity retention and 2nd in terms of overall MOS at ICASSP’s Signal Processing Grand Challenge (LIMMITS 2023).

Favicon imagearoraakshit.github.io

Guide on using NVIDIA RAPIDS on JFrog Artifactory

December 1, 2020

A guide written by Akshit Arora on how to use NVIDIA RAPIDS on JFrog Artifactory.

Favicon imagearoraakshit.github.io

Does Deep Knowledge Tracing Model Interactions among Skills?

January 1, 2018

Research paper exploring whether Deep Knowledge Tracing models interactions among skills.

Favicon imagearoraakshit.github.io

Interactive landslide simulator: a tool for landslide risk assessment and communication

January 1, 2016

Paper on an interactive landslide simulator designed as a tool for landslide risk assessment and communication. This work was partially supported by Thapar University and IIT Mandi.

Favicon imagearoraakshit.github.io

Hobbies

Enjoys photography and documents his attempts at capturing meteor showers. github+1