Karan Sindwani

Karan Sindwani

Senior Applied Scientist

Amazon Web Services

Karan Sindwani is a Senior Applied Scientist at AWS, with a decade of experience in machine learning and applied AI. His journey began in 2014 with his first academic paper, and since then, he has worked across a range of domains—from recommender systems and conversational agents at an AI startup , to cutting-edge computer vision research during his MS in Data Science at Columbia University, where he specialized in image inpainting. Since joining Amazon in 2020, Karan has played a key role in the launch of AWS Panorama, enhancing Amazon Personalize with graph-based recommendation systems.

In recent years, Karan has focused on generative AI, with a particular emphasis on fine-tuning small language models (SLMs) for vision-centric tasks. His portfolio includes pioneering work in zero-shot product placement in videos and contributions to multimodal understanding systems. A scientist at heart and a builder in practice, Karan bridges research and deployment to deliver scalable, intelligent systems that drive real-world impact.

In this session, I will share practical insights from actual production deployments of GenAI applications across multiple industries. Drawing from my experience with AWS, I will demonstrate:

1. Building a Customer Service Solution for Production - You will learn how we:

  • Implemented multilingual support through fine-tuned speech models
  • Achieved high throughput through strategic infrastructure scaling
  • Cut latency using NVIDIA TensorRT-LLM compilation
  • Deployed custom containers with Triton Inference Server

2. Automated Cricket Scene Analysis with Vision Language Models - I will show how we:

  • Reduced 45-50 minutes of manual processing per game to automated analysis
  • Built models to identify replays, bowler run-ups, and scorecard data
  • Developed resolution-agnostic, font-adaptive models for varying broadcast qualities
  • Optimized performance through Lora and Hyperpod fine-tuning

3. Gen AI-based Data Analyst - I will demonstrate how we:

  • Transform natural language queries into SQL and visualization code
  • Build conversational analytics assistants for complex data exploration
  • Set up secure database connections with proper authentication
  • Generate optimized data visualizations through LLM-powered code
  • Enable business users to perform sophisticated analytics without SQL knowledge
Read More

Managing and scaling ML workloads have never been a bigger challenge in the past. Data scientists are looking for collaboration, building, training, and re-iterating thousands of AI experiments. On the flip side ML engineers are looking for distributed training, artifact management, and automated deployment for high performance

Read More

Managing and scaling ML workloads have never been a bigger challenge in the past. Data scientists are looking for collaboration, building, training, and re-iterating thousands of AI experiments. On the flip side ML engineers are looking for distributed training, artifact management, and automated deployment for high performance

Read More
Book tickets View all speakers

We use cookies essential for this site to function well. Please click to help us improve its usefulness with additional cookies. Learn about our use of cookies in our Privacy Policy & Cookies Policy.

Show details