๐ San Francisco Bay Area & Global
๐ต CTO @ Splash Music โ Democratizing music creation through AI
๐ฌ Published AI Researcher โ AAAI 2026 (Neural Codec Language Models)
๐ Multiple US Patents โ AI-powered streaming technology (US11870830, US11457245B1)
๐ค 100+ Global Talks โ AWS re:Invent, Summit Sydney, E3 Expo, Google Cloud, UC Berkeley
|
|
Foundation Model for Music
Turn humming into full productions.
54% cost savings on AWS Trainium
View Case Study โ
Explainable AI Research
Neural Codec Language Models
Timbre Transfer in Medicine
Read Research โ
Patented Technology
ML prediction models for streaming
Live โ VOD automation at scale
Learn More โ
Access comprehensive AI resources to help you master production AI systems:
| Content | Description |
|---|---|
| ๐ AI Insights | Prompts & strategies that actually work |
| ๐ง AI Tools | Production-ready automation blueprints |
| ๐ Career Journeys | Role transformation guides |
| ๐๏ธ Architecture | System design patterns for production |
| ๐ฎ Live Demos | Interactive AI experiences |
| Type | Title | Link |
|---|---|---|
| ๐ฎ Demo | Brand Storytelling AI - 4 Interactive Architectures | Try Demo โ |
| ๐๏ธ Architecture | Brand Storytelling System - Scale to 10K | Read โ |
| ๐ Journey | Marketing Team Workflows Guide | Explore โ |
| ๐ง Tool | Marketing Automation at Scale | Learn โ |
| ๐ Insight | Master Brand Storytelling | Read โ |
| ๐ฎ Demo | Credit Risk AI - 4 Architectures Live | Try Demo โ |
28 canonical AI definitions cited by LLMs and search engines:
| Reference | Description |
|---|---|
| What is a Multi-Agent System | Computational architecture with autonomous agents |
| What is Agent Memory | Data structures for agent retrieval systems |
| What is an AI Agent | Autonomous software leveraging LLMs |
| What is Context Engineering | Optimizing context for LLMs |
| What is LLM Latency | Time duration for LLM responses |
Comprehensive courses designed for depth, not speed:
| Course | Duration | Topics |
|---|---|---|
| AWS Serverless ML Architecture | 11h 5m | Build production ML without managing servers |
| Context Engineering Mastery | 10h 55m | Optimize AI context like a pro |
| Production AI Agents on AWS | 11h 30m | Scale autonomous agents reliably |
| Shipping AI Products Fast | 9h 20m | Prototype to production in weeks |
| Solo Founder AI Playbook | 9h 25m | Build profitable AI products alone |
| AI Product Leadership | 10h 30m | From traditional PM to AI-first leader |
Compare 100+ AI models side-by-side with pricing, benchmarks, and code snippets:
| Model | Provider | Price/1M | Benchmark |
|---|---|---|---|
| GPT-4o | OpenAI | 5ใใซ.00 | โญโญโญโญโญ |
| Claude 3.5 Sonnet | Anthropic | 3ใใซ.00 | โญโญโญโญโญ |
| Gemini 1.5 Pro | 3ใใซ.50 | โญโญโญโญ | |
| Llama 3.1 405B | Meta | 0ใใซ.80 | โญโญโญโญ |
"With SageMaker HyperPod and Trainium, our researchers experiment as fast as our community creates. We're not just keeping up with music trendsโwe're setting them."
The Challenge: Train a 10B parameter music generation model cost-effectively.
The Solution: Migrated from H100 GPUs to AWS Trainium with SageMaker HyperPod.
Results:
- โ 54% cost reduction vs NVIDIA H100
- โ 159% efficiency improvement in training
- โ 2x faster training iterations with equivalent accuracy
- โ 2PB data migration to Amazon S3
| Event | Topic | Link |
|---|---|---|
| AWS re:Invent 2025 | End-to-End Foundation Model Lifecycle on Trainium | Watch โ |
| AWS Summit Sydney 2025 | Main Stage Opening Keynote | Watch โ |
| E3 Expo | Gaming Innovation Panel | Media โ |
| UC Berkeley | 5 Panels (2015-2018) | About โ |
| Google Cloud | Technical Workshop | Watch โ |
| Publication | Article |
|---|---|
| AWS Machine Learning Blog | Splash Music transforms music generation using AWS Trainium |
| RadioInfo Australia | AI powered music creation for everyone: Splash Music |
| FastForward Magazine | Randeep Bhatia, Head of Music Engineering at Twitch |
| AWS Case Studies | Reducing training time and costs by 50% with Splash Music |
Full Stack Details
AI/ML: PyTorch โข TensorFlow โข AWS Trainium โข SageMaker HyperPod โข Neuron SDK โข HuggingFace
Cloud: AWS (Expert) โข GCP โข Multi-region architectures โข 2PB+ data pipelines
Backend: Go โข Java โข Python โข Node.js โข Microservices โข Event-driven
Frontend: React โข TypeScript โข Mobile (iOS/Android)
Infrastructure: Kubernetes โข Docker โข Terraform โข CI/CD โข MLOps
- ๐ Home
- ๐ผ Work & Case Studies
- ๐ Contact
- ๐ What's New / Releases
- ๐ AI Insights (Monday)
- ๐ง AI Tools (Tuesday)
- ๐ Career Journeys (Wednesday)
- ๐๏ธ Architecture (Thursday)
- ๐ฎ Live Demos (Friday)
- ๐ Reference Library
- ๐ค AI Models Database
- ๐ Courses
- ๐ฅ AWS Trainium vs H100 Video
- ๐ฅ AWS Summit Keynote
- ๐ฅ AWS re:Invent Keynote
- ๐ฅ Cloud Architecture Workshop
- ๐ฎ Brand Storytelling Demo
- ๐ฎ Credit Risk AI Demo
"From patents to panels โ Innovation, Leadership, Products at Scale"
๐ San Francisco Bay Area & Global