Tushar is available for hire

Tushar Verma

Verified Expert in Engineering

AI and ML Engineer and Developer

Bengaluru, Karnataka, India

Toptal member since December 2, 2025

Expertise

Machine Learning Artificial Intelligence NLP Deep Learning Python PyTorch Docker NumPy LLM Linux Git

Bio

Tushar is an AI and ML engineer with four years of experience in developing natural language processing (NLP) and automatic speech recognition (ASR) systems, as well as model training, optimization, and deployment on GPUs. He works with Python, PyTorch, scikit-learn, Docker, and Triton, and manages the whole workflow from data to production. Tushar's work encompasses building, fine-tuning, and serving models at scale.

Portfolio

Convin

Python 3, PyTorch, NVIDIA Triton, Docker, FastAPI, Scikit-learn, NumPy, Pandas...

Experience

Linux - 4 years
Pandas - 4 years
Docker - 4 years
NumPy - 4 years
Python 3 - 4 years
PyTorch - 4 years
Scikit-learn - 2 years
NVIDIA Triton - 2 years

Preferred Environment

Python 3, PyTorch, NVIDIA Triton, Docker, NumPy, Pandas, Scikit-learn, FastAPI

The most amazing...

...project I've led was a multilingual code-mix ASR system, taken from data collection to training and full GPU deployment in production.

Work Experience

ML Engineer

2022 - 2025

Convin

Published Interspeech 2023 paper called "ASR for Low-resource and Multilingual Noisy Code-mixed Speech," proposing a novel Wav2Vec 2.0 training strategy for robust multilingual transcription in noisy conditions.
Trained and evaluated ASR models, including Wav2Vec 2.0, Whisper, and WavLM, achieving 15% best-case WER and an average of 21% WER on in-house noisy and code-mixed datasets.
Built inference infrastructure using NVIDIA Triton Inference Server, accelerating GPU-based deployments for production models.

Technologies: Python 3, PyTorch, NVIDIA Triton, Docker, FastAPI, Scikit-learn, NumPy, Pandas, Machine Learning, Product Development, Artificial Intelligence (AI), Data Science, Python, Natural Language Processing (NLP), Deep Learning, Text Classification, AI Consulting, Large Language Models (LLMs)

Experience

Interspeech 2023 | ASR for Low-resource and Multilingual Noisy Code-mixed Speech

Published a research paper proposing a novel Wav2Vec 2.0 training strategy designed to improve multilingual transcription in noisy and code-mixed environments. The work focused on low-resource scenarios, demonstrating substantial gains in robustness and accuracy across diverse acoustic conditions.

This project involved dataset preparation, experimentation, model optimization, and extensive evaluation, resulting in a successful peer-reviewed publication at Interspeech 2023.

Skills

Libraries/APIs

PyTorch, NumPy, Pandas, Scikit-learn

Tools

Git

Languages

Python 3, Python

Platforms

Docker, Linux

Other

Machine Learning, Artificial Intelligence (AI), Data Science, Natural Language Processing (NLP), Deep Learning, NVIDIA Triton, Product Development, Text Classification, AI Consulting, Large Language Models (LLMs), FastAPI, Hugging Face

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring