https://www.yashchudasama.com/ Recent content on Yash Chudasama Hugo en Copyright © 2026, Yash Chudasama; all rights reserved. Wed, 15 Apr 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/tech/sfu-vs-mcu-architecture/ Tue, 12 Aug 2025 00:00:00 +0000 https://www.yashchudasama.com/blog/tech/sfu-vs-mcu-architecture/ The first architecture decision in any real-time video system is the one that defines everything after it: SFU or MCU? Get it wrong and you’re rebuilding six months later. Get it right and your system scales naturally. Having built media infrastructure that handles thousands of concurrent sessions, I want to share the decision framework I use — not the textbook version, but the one shaped by production pain. The Fundamentals Selective Forwarding Unit (SFU) An SFU receives media streams from each participant and forwards them to everyone else. https://www.yashchudasama.com/blog/ai/voice-ai-latency/ Tue, 18 Nov 2025 00:00:00 +0000 https://www.yashchudasama.com/blog/ai/voice-ai-latency/ Voice AI has a problem that text-based AI doesn’t: time pressure. When a human asks a voice agent a question, they expect a response in the same cadence as a human conversation. Not in 2 seconds. Not in 1 second. In the time it takes a human to inhale before responding — roughly 200-500ms. Miss that window and the user perceives the system as broken, stupid, or both. Hit it consistently and the interaction feels magical. https://www.yashchudasama.com/blog/ai/ai-agents-autonomous-systems/ Sun, 15 Feb 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/ai/ai-agents-autonomous-systems/ The era of simple chatbots is over. AI agents — autonomous systems that can reason, plan, and execute complex tasks — are fundamentally reshaping how we build and interact with software. As someone working at the intersection of AI and real-time communication, I’ve seen firsthand how this shift is transforming the technology landscape. What Are AI Agents? AI agents are systems that go beyond simple prompt-response interactions. They can: Reason about complex problems and break them into subtasks Plan multi-step approaches to achieve goals Execute actions using tools, APIs, and external systems Learn from feedback and adapt their strategies Collaborate with other agents in multi-agent workflows Unlike traditional AI applications that respond to single queries, agents maintain context, make decisions, and take actions autonomously. https://www.yashchudasama.com/blog/tech/edge-computing-future/ Tue, 20 Jan 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/tech/edge-computing-future/ The cloud revolution centralized computing. Now edge computing is decentralizing it again — but smarter. Having built real-time communication infrastructure at scale, I’ve experienced firsthand why pushing computation closer to users isn’t just an optimization; it’s a fundamental architectural shift. Why Edge Computing Matters Now Three converging trends are making edge computing essential: 1. The Latency Problem Cloud data centers are fast, but physics imposes limits. A round trip from Mumbai to a US-East data center takes 200-300ms. https://www.yashchudasama.com/blog/tech/cloud-media-workloads/ Wed, 15 Apr 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/tech/cloud-media-workloads/ Cloud providers are optimized for request-response workloads. Real-time media — WebRTC, voice AI, live streaming — breaks their assumptions in fundamental ways. Standard cloud architecture patterns (load balancers, auto-scaling groups, stateless microservices) don’t work for media. Here’s what does, and why the default advice from cloud providers will cost you latency, money, or both. Why Media Workloads Are Different A typical web request hits a load balancer, gets routed to any available instance, processes for 50-200ms, and returns. https://www.yashchudasama.com/blog/ai/realtime-ai-pipelines/ Wed, 25 Mar 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/ai/realtime-ai-pipelines/ The most interesting AI systems in 2026 aren’t chatbots. They’re voice agents — AI that listens, thinks, and speaks in real-time within live communication flows. Building one that actually works in production requires solving problems at the intersection of WebRTC, speech processing, and language models. This is the architecture guide I wish existed when I started building these systems. The Reference Architecture A production real-time AI voice pipeline has five layers: https://www.yashchudasama.com/blog/tech/webrtc-production-debugging/ Tue, 10 Mar 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/tech/webrtc-production-debugging/ WebRTC documentation teaches you the happy path. Production teaches you everything else. After years of building real-time communication infrastructure, I’ve compiled the debugging scenarios that consumed the most engineering hours — the ones that don’t appear in tutorials but appear in your error logs at 3 AM. 1. The Symmetric NAT Problem What the docs say “Use STUN to discover your public IP, then use it to establish a peer connection. https://www.yashchudasama.com/blog/ai/ai-and-human-creativity/ Sun, 01 Mar 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/ai/ai-and-human-creativity/ Every transformative technology triggers the same fear: will it replace us? The printing press threatened scribes. Photography threatened painters. Calculators threatened mathematicians. Now artificial intelligence threatens… everyone? As someone who works with AI daily and thinks deeply about its philosophical implications, I believe this framing fundamentally misunderstands both AI and human creativity. The Nature of Human Creativity Before we can assess AI’s impact on creativity, we need to understand what creativity actually is. https://www.yashchudasama.com/blog/tech/wasm-beyond-browser/ Sat, 28 Feb 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/tech/wasm-beyond-browser/ WebAssembly (Wasm) was created to run code at near-native speed in web browsers. But its most transformative impact may be outside the browser entirely. Wasm is becoming the universal runtime — a portable, secure, fast execution environment for everything from server-side applications to IoT devices to plugin systems. Why WebAssembly is Breaking Out Wasm has properties that make it uniquely suited as a universal runtime: Near-Native Performance Wasm executes at speeds within 10-20% of native code. https://www.yashchudasama.com/blog/ai/rag-building-knowledge-systems/ Mon, 05 Jan 2026 00:00:00 +0000 https://www.yashchudasama.com/blog/ai/rag-building-knowledge-systems/ Large Language Models know a lot, but they don’t know your data. Retrieval-Augmented Generation (RAG) bridges this gap — enabling AI systems that combine the reasoning power of LLMs with the specificity of your own knowledge base. Having built several RAG-powered applications, I want to share practical insights on building systems that actually work. What is RAG? RAG is an architecture pattern that enhances LLM responses by retrieving relevant information from external sources before generating answers. https://www.yashchudasama.com/blog/ai/google-a2a-protocol/ Sun, 04 May 2025 00:00:00 +0000 https://www.yashchudasama.com/blog/ai/google-a2a-protocol/ Google’s Agent-to-Agent (A2A) protocol represents a groundbreaking advancement in how AI agents communicate and collaborate with each other. This protocol is designed to enable seamless, structured, and efficient interactions between multiple AI agents, opening up new possibilities for complex problem-solving and automation. What is the A2A Protocol? The Agent-to-Agent protocol is a standardized framework developed by Google that enables AI agents to: Communicate effectively with each other Share information and capabilities Coordinate actions and decisions Maintain context across interactions Execute complex multi-agent workflows Key Features of A2A The protocol incorporates several innovative features: https://www.yashchudasama.com/blog/ai/llm-revolution/ Sun, 03 Nov 2024 00:00:00 +0000 https://www.yashchudasama.com/blog/ai/llm-revolution/ Large Language Models (LLMs) have emerged as one of the most transformative technologies of our time. As someone working at the intersection of technology and philosophy, I find the implications of LLMs particularly fascinating. What are Large Language Models? LLMs are AI systems trained on vast amounts of text data to understand and generate human-like language. They represent a significant leap in natural language processing capabilities, enabling machines to: Understand context and nuance Generate coherent and contextually relevant text Perform complex language tasks Learn from examples and adapt to new situations The Technical Foundation At their core, LLMs are built on: https://www.yashchudasama.com/blog/tech/webrtc-revolution/ Sun, 27 Oct 2024 00:00:00 +0000 https://www.yashchudasama.com/blog/tech/webrtc-revolution/ WebRTC (Web Real-Time Communication) has transformed how we think about real-time communication on the web. Having spent years building real-time communication infrastructure, I’ve witnessed firsthand how this technology has revolutionized digital interactions. What Makes WebRTC Special? WebRTC is more than just another communication protocol. It’s a collection of protocols, standards, and JavaScript APIs that enable real-time communication directly between browsers. The key features that make it revolutionary include: Peer-to-Peer Communication: Direct browser-to-browser connection Low Latency: Near real-time data transfer High-Quality Media: Support for HD video and audio Security: Built-in encryption and security features Cross-Platform: Works across different browsers and devices The Technical Magic At its core, WebRTC uses three main APIs: https://www.yashchudasama.com/contact/ Sun, 28 Jul 2024 00:00:00 +0000 https://www.yashchudasama.com/contact/ Problems I Think About Scaling WebRTC infrastructure — SFU architecture, cascading, multi-region deployment Voice AI latency — building pipelines where the response starts before the user finishes thinking Cloud architecture for media — why standard patterns break and what to use instead Integrating AI into live communication — real-time STT, LLM, and TTS within WebRTC sessions If you’re working on something in these areas, I’d enjoy talking about it. Get in Touch Email The best way to reach me for anything substantive: https://www.yashchudasama.com/about/ Sat, 27 Jul 2024 00:00:00 +0000 https://www.yashchudasama.com/about/ What I Do I design and build real-time communication systems and AI infrastructure. My work spans WebRTC, voice AI pipelines, media server architecture, and cloud-native platforms operating at serious scale. I’m drawn to the hardest problems in this space — the ones where latency budgets are measured in milliseconds, where a single architecture decision determines whether a system can handle 10 users or 10,000, and where AI needs to work in real-time, not batch. https://www.yashchudasama.com/archives/ Tue, 28 May 2019 00:00:00 +0000 https://www.yashchudasama.com/archives/