Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
LLM Nutrition Pipeline Architecture
Explore the backend architecture of an LLM nutrition engine, detailing a multi-stage pipeline for parsing, database routing, and scoring real-world meal data.
In this session, I’ll walk through the real backend architecture behind an LLM-powered nutrition engine that converts natural language meal descriptions into structured nutrition data. Rather than treating the LLM as a black box, this system constrains it within a multi-stage pipeline: parsing, classification, database routing, scoring, and caching.
I’ll share whats been working for me so far, and all the things that didn’t when integrating LLMs with real food databases, handling ambiguous user input, and designing systems that remain explainable despite relying on probabilistic models. The focus is not on prompt tricks or hype, but on system design, tradeoffs, and lessons learned from shipping an AI-assisted backend, coded by AI.
Related projects
Developing AI Solutions for Busy Parents
Toronto
A demo of a custom email summarizer that extracts key details and dates, with notebooks covering model selection,…
Human-ish: A Mirror for the AI Age
Toronto
This talk introduces Human-ish, a tool to detect AI-written LinkedIn posts, exploring the feedback loop between AI content…
Bringing STEM to Life: Automating Manim Visualizations with LLMs
Toronto
Learn how EduWiz uses LangChain and OpenAI to generate Manim animations on demand, creating interactive, automatically rendered tutorial…
Autobots - Loblaws Supply Chain agent
Toronto
This talk covers Autobots, an AI agent managing Loblaws supply chain exceptions by autonomously emailing vendors, processing replies,…
Observability that drives true ROI
Toronto
This talk introduces AI managers that monitor and summarize agent behavior, moving beyond dashboards to quickly show what…
What Happens When You Let AI Build a Game?
Toronto
A live demonstration of building two Three.js/Nuxt browser games in two weeks using AI, covering prompt strategy, code…