Deepesh Agrawal (@Deepesh76780)

deepesh76780.hashnode.dev5d ago · 5 min read

One worker per agent was the wrong shape: building a shared pool for voice agents

Early on, every voice agent in the platform I was building ran as its own process on one EC2 box, with its own Start button. This post is about what that cost, and how a call travels through the serve

0

deepesh76780.hashnode.devJul 18 · 26 min read

The Anatomy of a Voice Agent: What Happens in the 550 Milliseconds After You Stop Talking.

A line-by-line tour of LiveKit's realtime stack - from the SFU that moves your packets to the four-layer language model that decides you've finished your sentence. Everything below was read directly f

0

deepesh76780.hashnode.devSep 15, 2025 · 9 min read

JWT: A simple step-by-step deep dive (so you actually understand it)

Skip to contentJSON Web Tokens explained from first principles, with manual signing, secure patterns, common pitfalls and ready-to-use code. Why this guide? Lots of “JWT guides” either copy the specification or show a tiny code snippet. This one exp...

0

deepesh76780.hashnode.devSep 11, 2025 · 5 min read

Designing a Scalable Rate Limiting Architecture — Lessons from an Interview

A few days ago, during a system design interview, I was asked a simple-looking but tricky question: “How would you design a scalable rate-limiting algorithm for millions of API requests per second?” At first glance, rate limiting sounds straightfor...

0

deepesh76780.hashnode.devSep 9, 2025 · 5 min read

Understanding WebRTC for Voice Bots — A Desi Developer's Guide

By Deepesh Agrawal 👋 Why I’m Writing This Three weeks ago, I didn’t even know what WebRTC was. I was happily building voice chatbots using the "normal" way: Record audio → Send to API → Wait → Get response → Play back. Worked okay... until I tried ...

0

Deepesh Agrawal

About

Available for

Deepesh Agrawal's blogs

One worker per agent was the wrong shape: building a shared pool for voice agents

The Anatomy of a Voice Agent: What Happens in the 550 Milliseconds After You Stop Talking.

JWT: A simple step-by-step deep dive (so you actually understand it)

Designing a Scalable Rate Limiting Architecture — Lessons from an Interview

Understanding WebRTC for Voice Bots — A Desi Developer's Guide

Deepesh Agrawal

About

Available for

Deepesh Agrawal's blogs

One worker per agent was the wrong shape: building a shared pool for voice agents

The Anatomy of a Voice Agent: What Happens in the 550 Milliseconds After You Stop Talking.

JWT: A simple step-by-step deep dive (so you actually understand it)

Designing a Scalable Rate Limiting Architecture — Lessons from an Interview

Understanding WebRTC for Voice Bots — A Desi Developer's Guide

Deepesh Agrawal

About

Available for

Deepesh Agrawal's blogs

One worker per agent was the wrong shape: building a shared pool for voice agents

The Anatomy of a Voice Agent: What Happens in the 550 Milliseconds After You Stop Talking.

JWT: A simple step-by-step deep dive (so you actually understand it)

Designing a Scalable Rate Limiting Architecture — Lessons from an Interview

Understanding WebRTC for Voice Bots — A Desi Developer's Guide

Search Hashnode

Deepesh Agrawal

About

Available for

Deepesh Agrawal's blogs

One worker per agent was the wrong shape: building a shared pool for voice agents

The Anatomy of a Voice Agent: What Happens in the 550 Milliseconds After You Stop Talking.

JWT: A simple step-by-step deep dive (so you actually understand it)

Designing a Scalable Rate Limiting Architecture — Lessons from an Interview

Understanding WebRTC for Voice Bots — A Desi Developer's Guide