top of page

Hello

Here's who I am & what I do

Engineering leader with 20+ years of experience building production-grade AI systems spanning machine learning, Generative AI/LLMs, Agentic AI, data science, search, analytics, and full-stack platforms.
 

I joined SoundHound AI in 2014 when it was a startup, and helped scale it through its journey to becoming a public company in 2022 (SOUN).

I work on next-generation conversational AI — bringing natural voice experiences to phones, cars, TVs, speakers, coffee machines & every other part of the emerging 'connected' world!

profile_photo.jpg

Pranav Singh

VP of Engineering  Building LLMs, Agentic AI

  • LinkedIn
  • X

Key Projects

When Vision Meets Voice: Elevating Enterprise AI Through True Multimodal Intelligence.

This domain lets users of its in-vehicle voice AI platform access the car manual using natural speech.

Proprietary Machine Learning technology named CaiNet (Conversational AI Network) to better understand queries and ensure fast, accurate, and appropriate responses.

The Wanna-Be-So™ series is designed to make everyday life effortless with simple, intuitive, and highly practical apps.

Category-defining voice AI technology that enables restaurant employees from both the front and back-of-house to ask a custom-trained voice assistant critical questions – and get immediate responses – completely hands-free.

A single, easy to use product, built natively on Hadoop to transform raw data into business insight in minutes, without the need to learn complex products or rely only on highly skilled resources.

State-of-the-art technologies,sophisticated algorithms and a user-friendly interface, Nextag enables smart shoppers to quickly compare the internet's lowest prices and make the perfect purchase – every time

Patents 

7 Registered Patents

1

Bidirectional probabilistic natural language rewriting and selection

Improves speech recognition accuracy by automatically rewriting low-confidence inputs using forward and backward language models, grammar scoring, and graph-based optimization.

Date Issued

March 24, 2020

2

Machine learning system for digital assistants

Uses sequence-to-sequence models and embedding-based clustering to transform user queries into canonical representations, enabling scalable training and improved intent normalization.

Date Issued

August 20, 2024

3

Natural language module store

A platform for publishing, discovering, testing, and composing reusable natural language modules to accelerate development of conversational systems.

Date Issued

October 16, 2018

4

Neural sentence generator for virtual assistants

Generates realistic example user phrases using fine-tuned language models and domain data, speeding assistant configuration and improving intent coverage.

Date Issued

5

Predicting human behavior by machine learning of natural language interpretations

Creates “thought maps” by combining natural language understanding with location and contextual signals to analyze collective intent, predict behavior, and enable location-aware applications.

Date Issued

May 21, 2019

6

System and method for detection and correction of a query

Identifies speech and language errors using query similarity, semantic embeddings, and interaction signals, enabling automated data labeling and continuous model improvement.

Date Issued

March 1, 2022

7

Token confidence scores for automatic speech recognition

Automatically detects and replaces low-confidence transcription tokens using neural confidence scoring, while exposing confidence signals to support labeling and vocabulary enhancement.

Date Issued

February 11, 2025

Publication

Cyber Security and Global Information Assurance: Threat Analysis and Response Solutions

IGI Global · Jan 1, 2009

© 2026 by Pranav Singh.
 

Follow

  • LinkedIn
  • X
bottom of page