WebAI Badge Studio: Espruino at Google WebAI Summit
How a 45-minute portrait queue turned into a live USB-C hardware hack, a bit-exact 96,000-byte 2bpp BWYR codec, surgical .bootcde firmware patching, and an open-source WebAI Badge …
Racecraft: Fine-Tuning Gemma 4 with DPO for Sub-150ms On-Device Coaching at 130 MPH
Read full post →
The Throughput Trap: Benchmarking vLLM on OpenXLA and the Reality of Production LLM Serving
vLLM Systems · DevLab 2026, Deep Dive I was recently invited by the Google TPU team to speak at the OpenXLA Summer DevLab 2026 . This post breaks down our deep-dive evaluation of …
Splitting the Brain to Beat the Clock
How a "brake!" lands in 5 milliseconds while a cloud model thinks for five seconds — in the same app, on the same frame, without ever colliding.
Teaching the Coach to Read the Driver
The best instructors don't coach the car. They coach you , and they figure out who you are in about two laps. Here's how we taught software to do the same.
The 800-Millisecond Problem
Why I stopped trusting every " AI race coach " I'd ever tried , and what it would take to build one I'd actually listen to at 130 km/h.
Racecraft (Project Koru) · Prologue — The Origin Story
The origin story of Project Koru — building an AI Race Coach at Sonoma Raceway with Google Developer Experts, Antigravity, and Gemini.
The Blind Spot Horizon: Why Your AI Benchmarks Are Lying to You
"If you measure the wrong thing, you optimize for the wrong thing. And when dealing with frontier, multimodal models, the wrong thing might just break your entire system without …
My Friend's MRI Didn't Come with a Manual, So I Built One with AI
The other was a flimsy paper sleeve containing a CD-ROM. This, we were told, held the actual images from her MRI scan. The ground truth. And we couldn't even look at it.
The Hidden Performance Trap in Causal Ring Attention
This post is a walkthrough of ring-flash-jax , a small JAX project that explores this problem and implements the fix. We'll add the four things you actually need before Ring …
Unboxing Gemma: High-Throughput Sparse Autoencoder (SAE) Training on Google Cloud TPU v5e
A blog post about unboxing gemma: high-throughput sparse autoencoder (sae) training on google cloud tpu v5e
The Gemma 4 E2B Fine-Tuning Cookbook
A complete, opinionated recipe for adapting Gemma 4 E2B to your domain — from multimodal dataset construction and QLoRA configuration through training loop debugging, evaluation, …
Building a Multimodal Document Intelligence Pipeline with Gemma 4 E2B
A production-grade walkthrough — PDF ingestion, OCR, audio transcription, structured reasoning, and agentic tool calling — all on a 2.3B parameter model that fits on a single …
Write Once, Scale Everywhere
End-to-End Gemma 2B LoRA Fine-Tuning and Serving on GPU & TPU If you have ever prototyped a Large Language Model LLM on your local GPU and then s......
Visualizing the Invisible: How Nano Banana 2 Turns Dense Science into Stunning Art
Disclaimer: As a Google Developer Expert (GDE), I was incredibly fortunate to be invited by Google DeepMind to test these models internally before their public release. The …
Fine-Tuning at the Redline: Architecting a Trustable AI Racing Coach
" AI in a Jupyter notebook is safe. AI on a race track—at 150 mph—is a different story." - Ajeet Mirwani
A Split‑Brain Neuro‑Symbolic Training Method for High‑Velocity Autonomous Coaching from Telemetry
A blog post about a split‑brain neuro‑symbolic training method for high‑velocity autonomous coaching from telemetry
The Context Trap: Why Scaling Laws Can’t Break the Ceiling of Uncertainty
We are currently living through the “ Post-Reasoning ” phase of the AI hype cycle. By now, models like Gemini 2.0 and the latest iterations of Gemini have normalized the idea that …
Building Production AI Systems with Free Cloud GPUs
If you've been following my work on LLM-assisted code generation and AI reasoning , you know I'm always looking for ways to democratize AI development . During my recent work on …
🎤 SpeakWise: Build an AI Public Speaking Coach with Gemma 3n
Project by: Rabimba Showcase for: Google Developer Expert (GDE) AI Sprint
From Models to Agents: Shipping Enterprise AI Faster with Google’s MCP Toolbox & Agent Development Kit
This article is an expanded write-up of the talk I recently delivered as a Google Developer Expert during a talk in Denver. The full slide deck is embedded below for easy …
Deep Dive into the Google Agent Development Kit (ADK): Features and Code Examples
In our previous overview, we introduced the Google Agent Development Kit (ADK) as a powerful Python framework for building sophisticated AI agents. Now, let's dive deeper into some …
Build Smarter AI Agents Faster: Introducing the Google Agent Development Kit (ADK)
The world is buzzing about AI agents – intelligent entities that can understand goals, make plans, use tools, and interact with the world to get things done. But building truly …
My Google I/O 2024 Adventure: A GDE's Front-Row Seat to the Gemini Era
Hey tech enthusiasts! Rabimba Karanjai here, your friendly neighborhood Google Developer Expert (GDE), back from an exhilarating whirlwind tour of Google I/O 2024. Let me tell you, …
MovieBuff: Dive Deeper into Movies with Generative AI
MovieBuff: Dive Deeper into Movies Before You Watch MovieBuff: Dive Deeper into Movies Before You Watch Have you ever spent two hours watching a movie only to be disappointed? …
Curious case of Cisco AnyConnect and WSL2
One thing Covid has taught me is the importance of VPN. Also one other thing COVID has taught me while I work from home is that your Windows Machine can be brilliant as long as …
Immersive Payment: Second Update
This is the second update to the Immersive Payment project update under the Grant For Web Fellowship. You can read the last report here: …
Immersive Payment: First Update
It has been more than 6 months now that I have had the Immersive Payments, Spark Grant. Since then I have worked on realizing the scope of the project and finally, I am here to …
Road to the Grant : Immersive Payment
This is a news close to my heart. My project "Immersive Payment" has been awarded with Grant for the Web , and I will be focusing on using Web Monetization to enable Micro …
Attack of the Clones (Part 1): Drishti and vMeet
Background : India recently banned certain Chinese Apps from being used and also with that launched a campaign of being vocal for local . Essentially promoting industries and app …
Handling Web Server file permission: The Last Guide
If you have to work with a LAMP stack one of the first things you do is setting up permission for your user and web server so that you can deploy files correctly and securely and …
Help finding a cure for COVID19: Set up Folding@Home
COVID-19 has been already assigned as a pandemic worldwide now. It is now a global problem affecting all of our day to day lives. Almost all the methods of combating the virus at …
Blocking Communication - Breaking Out
I disapprove of what you say, but I will defend to the death your right to say it If you have been following my previous post you will realize how quickly a state actor can just …
Blocking Communication - A perspective from India : The Beginning
Update: If you want to know how to circumvent this kind of blocking. head over to my second part of the blog post. If liberty means anything at all, it means the right to tell …
Visualizing large scale Uber Movement Data
New York's cab data visualization from Uber's Engineering blog Last month one of my acquaintances in LinkedIn pointed me to a very interesting dataset. Uber's Movement Dataset. It …