I used to think that I love coding, but in the last year, I came to realize that what I love more is building - creating something useful and beautiful. Last year this time, my coding workflow was to fire up VS Code with Claude in the browser or GitHub Copilot in ‘Ask’ mode and brainstorm with a model, review solutions suggested by the LLM, copy code into the editor, test and deploy🚀. This was fun in the beginning, but soon the context-switching became tedious and broke the flow of building.
In our previous post, Mobile On-device AI: Smarter Faster Private Apps, we explored the fundamentals of running AI locally on mobile devices. Now, it’s time to get hands-on and see this technology in action!
This practical guide walks you through implementing mobile on-device AI using Google’s powerful Gemma model family, including the cutting-edge Gemma 3n. You’ll learn to deploy these models across iOS, Android, and web platforms using industry-standard frameworks.
What You’ll Learn Here
- Test various Gemma models, including Gemma 3n, in Google AI Studio.
- Run sample on-device AI applications on Android using tools like the Google AI Edge Gallery App and MediaPipe.
- Implement on-device Large Language Model (LLM) inference on iOS using MediaPipe.
- Explore how to run LLMs in mobile web browsers with JavaScript and MediaPipe.
- Gain practical experience deploying and interacting with Gemma models across different mobile platforms.
Prerequisites: Basic mobile development knowledge helpful but not required.
While cloud computing drives many AI breakthroughs, a parallel revolution is happening right in our hands - running LLMs locally on mobile devices. This emerging field, known as Mobile On-device AI, enables us to build more private, faster and smarter app experiences - especially as mobile devices become increasingly powerful. As a developer passionate about AI and mobile, I am fascinated by the convergence these two worlds and the possibilities it brings.
Every day we’re seeing fantastic advancements in AI, thanks to more data and powerful computers. This may make it seem like the future of AI is all about getting even more data and bigger computers. But I believe a critical and rapidly evolving piece of the puzzle is about bringing the Intelligence of Artificial Intelligence onto the devices where the data originates (eg: our phones, cameras, and IoT devices) and doing the “smarts” using their own computing capabilities. This is the essence of Edge AI, the topic we’ll explore in this post.
We know that RAG (Retrieval Augmented Generation) is a reliable mechanism to augment LLMs with up-to-date data and ground them on facts relevant to the context of the user query, thereby reducing hallucination. When set up properly, it works pretty well. Companies like Perplexity AI and enterprise applications use RAG extensively.
However, building a RAG pipeline on your own from scratch can be complex and high maintenance. You need to assemble your data sources, chunk the data, index it, generate embeddings, and store them in a vector database. At inference time, you need to generate an embedding of the user query using an embedding model, retrieve the relevant data from the indexed store and return a meaningful context-aware response to the user. On top of that, any change in the data source means that you need to re-index the data, re-generate embeddings, and update the store. Rinse and repeat.
I stopped using Chrome a long time ago due to privacy concerns, and I’ve been switching between Firefox, Edge, and occasionally Brave. But I never spent the time to fully compare these browsers and analyze which one best suits my needs. Today, I decided to look deeply into these browsers and ended up going down a rabbit hole of researching their privacy features.
Why is this important? Every time we browse a website, we are sharing bits of our digital life with companies we don’t know. These companies collect vast amounts of data about our browsing habits, preferences, and personal information. It is so cringy and annoying to see YouTube showing a ‘suggested video’ of the dental procedure that you searched about and are planning to get done. The browser you choose plays a significant role in how much of your data is collected and shared.
Today I tried something fun - built a Pomodoro timer app mostly by talking to AI instead of typing code myself. And guess what - there is a term for it - vibe coding, coined by Andrej Karpathy 😎.
I have done it a few times before, but this is the first time I am using it to build a full app. I wanted to create something that was useful and worked well, so I chose the Pomodoro timer. Here’s how it went and my key takeaways from this way of building products.
A few months ago, if you had told me I’d be deadlifting 155 lbs, I would have laughed and said, “No way!” But here I am, doing it—something I never thought possible.
In a previous post in my reimagine journey, I talked about reframing milestones to focus on consistency rather than arbitrary goals. At the time, I was working out on my own, trying to stay active with strength training and Zone 2 walks. I was making some progress, but it was small and I felt stuck. Looking back, I realize it wasn’t a lack of effort—it was a lack of structure, consistency, and progression.
Today I tried Claude Code, the new agentic coding tool announced by Anthropic this morning. Unlike other agentic tools, Claude Code is a CLI tool.
Claude has been my favorite AI coding partner so far. I use it via GitHub Copilot and as standalone through its web interface. I was curious to see how it works in CLI and decided to give it a try.
In this post, I share my first impressions of using Claude Code - how I set it up, what I loved about it, what I didn’t, and how it compares to other similar tools.
Of late, I have been building and learning about browser extensions for a few projects. It was surprising to learn that there are many ways to build UI for these extensions and the most interesting one was the Side Panel UI available to Chrome extensions. In this post, I will talk about the Chrome side panel, how to build one and the advantages and limitations of using it.
The Chrome side panel
It is a vertical panel that opens on either side of the browser window, providing quick access to your bookmarks, history, reading list, and Google Lens. It has a consistent layout and behavior across all websites. It was first introduced in Chrome in May 2023.