Machine Learning Engineer at Red Hat

Gaurav Goswami

7+ years building production AI systems — from LLM inference optimization and distributed training to diffusion models shipped on Samsung Galaxy devices.

7+
Years
176
Repos
1.2K+
Stars
1800+
DSA Solved

Transforming Ideas into AI Solutions

Building open-source AI infrastructure at scale

I am a Machine Learning Engineer at Red Hat with 7+ years of experience developing cutting-edge AI solutions, from mobile camera systems to large-scale model training infrastructure.

At Red Hat, I work on open-source AI infrastructure and large-scale model systems, contributing to the PyTorch ecosystem, developing scalable training workflows using Training Hub, and optimizing high-throughput inference with vLLM for large language models.

Passionate about both research and implementation, I've authored patents and conference papers while maintaining an active profile in competitive programming with 1800+ DSA problems solved.

Red Hat AI Infra
LLM & vLLM
PyTorch Ecosystem
Open Source
Gaurav Goswami

Professional Journey

7+ years of building production-grade AI systems

Red Hat

Machine Learning Engineer

Sep 2025 - Present
  • Open-source AI infrastructure and large-scale model systems
  • PyTorch ecosystem contributions and scalable training workflows (Training Hub)
  • High-throughput LLM inference optimization with vLLM
  • Infrastructure for training and serving foundation models

Samsung Research Institute, Bangalore

Computer Vision Lead Engineer

Nov 2021 - Aug 2025
  • SOTA low-light image/video enhancement with deep learning denoising
  • Commercialized Moiré Removal (Galaxy S23) and Deblurring (Galaxy S24)
  • Diffusion Models for image generation & restoration
  • Real-time mobile inference via QAT & pruning
  • Multiple Samsung Best Paper Award (SBPA) recognitions

Synergy Labs

Deep Learning Engineer

Jul 2018 - Oct 2021
  • Video analytics for traffic detection & people counting
  • ALPR system with anchor-based detector + multi-head OCR on Jetson TX2
  • GAN-based synthetic data generation and document OCR pipeline
  • Vehicle axle classification system for toll management

Tech Stack

LLMs & GenAI

vLLMtorchtuneLoRA / QLoRAKV Cache OptimizationDiffusion ModelsFlow Matching

AI Infrastructure

Distributed TrainingFSDPModel ServingTraining HubOpen Source

Deep Learning & Optimization

PyTorchONNXQuantization (QAT / GPTQ)PruningTensorFlow

Computer Vision

Image EnhancementDenoisingObject DetectionGenerative InpaintingOpenCV

Programming & Systems

PythonC++CUDALinuxGitDocker

Competitive Programming

1900+ problems solved across platforms

Academic Background

Mahatma Jyotiba Phule Rohilkhand University, Bareilly

B.Tech, Computer Science Engineering

2014 - 2018

Let's Work Together