I like building things
and writing about what I learn
Product manager, engineer, tinkerer, occasional writer. I spend most of my time building with AI, figuring out how things work, and shipping side projects on weekends.
Things I made
-
Describe any character or scene, and get a printable coloring page in seconds. Multi-turn chat refinement, age-appropriate output, and a showcase feature where kids share their finished art.
-
Upload your Gmail export and see your travel story on an interactive 3D globe. On-device AI extracts flight data from confirmation emails with zero server calls.
-
Self-learning agent that observes your market across 11 data sources, creates Instagram Reels & Stories, and improves every week. 8-tool reasoning engine with Fair Housing compliance built in.
Worth a read
How I Got OpenClaw Running on My ChatGPT Subscription After Hours of Bad Advice
A practical setup guide for self-hosting OpenClaw on a Hostinger VPS using your ChatGPT subscription via Codex OAuth, plus what I wish I'd known an hour sooner.
Published:FlightWrapped: Turning Gmail into a 3D Travel Story with On-Device AI
How I built FlightWrapped, a privacy-first flight visualizer that runs an LLM in your browser to extract flights from Gmail exports and render them on a 3D globe.
Published:GPT Car: How I Made ChatGPT Drive My Son’s Toy Car
A weekend project that turned ChatGPT into the driver of a toy car.
Published:Vibe Coding with Llama API
Building SkinSync: How I Vibe Coded a Makeup Recommendation App in One Weekend
Published:Have you wondered how a typical LLM is built?
I'm going to take just 5 minutes of your time to explain how Llama 3 was built
Published:Would Model Context Protocol be the catalyst for AI Agent Revolution?
Can the Model Context Protocol (MCP) facilitate broader interoperable context layers for LLMs, thereby paving the way for a revolution in AI agentic workflows?
Published:Text Streaming in AI Interfaces: Understanding Real-Time Content Generation
Streaming in AI Interfaces is important and it creates an illusion of natural communication
Published:LLM Prompt Routing: The Strategic Key to Scaling AI Applications
Prompt routing dynamically directs AI prompts to the right model based on cost, complexity, and task type. Learn key routing strategies for scaling AI apps efficiently.
Published:Building an Offline AI tool in your browser
Build a lightweight ChatGPT-like chatbot that runs entirely offline in Chrome using Gemini Nano — no servers, no API keys, just the browser.
Published:On Device AI
On-device AI presents a compelling opportunity for businesses to achieve cost savings alongside the well-established benefits of AI.
Published:
Recent Posts
Generative UI
User Interfaces (UI) in mobile apps and web apps are going through a `huge rethink` with the advent of LLM applications both in consumer and enterprise apps.
Published:Building AI Agents
Build a pair of AI agents with Microsoft AutoGen — one scrapes YC companies, the other analyzes their products. A practical walkthrough with code.
Published:Forcing a LLM to operate Safe
How to enforce LLM safety using LLM Guard to sanitize and filter prompts — essential for apps targeting children or sensitive audiences.
Published:How to run a LLM in your Laptop
LLMs can be quantized and be fit in your laptop's GPU. Really!
Published:The Oxymoron effect of AI
In the rapidly evolving field of Artificial Intelligence (AI), two key factors dominate the discourse: efficiency and cost.
Published:'App' less Phones are the future?
Could conversational AI replace siloed mobile apps? Exploring how a unified chatbot-style interface might streamline everything from banking to daily tasks.
Published: