
Andrew McNamara
@Drewch
Followers
1K
Following
2K
Media
41
Statuses
310
Building Sidekick @Shopify | Director of Machine Learning Engineering | Bringing AI to entrepreneurs | Bush Pilot | Maple Syrup Producer https://t.co/haSezvLkIm
Belwood, ON
Joined January 2010
I've been chatting a lot about evals for LLM systems and how to get away from "vibe evals" to statistically robust LLM as a Judge evaluations. Had an awesome chat with @andrewzigler from @DevInterrupted about @Shopify Sidekick and Vibe Entrepreneurship. 🧵.
3
4
24
RT @FoxEcomOfficial: This is #ShopifySidekick's answer after scanning our tested store, and it gives us details exactly what to do to faste….
0
1
0
Had a blast talking at @icmlconf 2025 this year in Vancouver talking about how you should be evaluating your agents in a robust manner. No more vibe testing!.
@Drewch In addition to the latest blog post, our talk at #ICML2025 is now live! 📺✨ Learn about the lessons & results behind building production-ready agentic systems:
0
0
7
A great prompt that I used recently to send an email/loyalty discount out to some customers who haven't ordered in a while: ."Show me my top 5 customers who make the biggest orders who haven't made an order in 3 months and draft an email to send them with a discount code for a.
2
0
13
RT @ShopifyEng: Crafting agentic systems requires smart architecture, rigorous evaluation, & relentless failure-proofing. 🏗️🧪.Director of M….
shopify.engineering
Learn how we evolved our AI assistant architecture and built robust evaluation frameworks for real-world deployment
0
6
0
@andrewzigler @DevInterrupted @Shopify You can also check out a talk I gave at ICML this year in Vancouver with @tempor1s "Building Production Ready Agentic Systems: Architecture, LLM-based Evaluations, and GRPO Training:
0
0
4
@andrewzigler @DevInterrupted @Shopify Here's a link to the @DevInterrupted podcast with @andrewzigler:
0
0
3
Looking for internship working on cutting-edge agentic systems? Apply now!.
🚀 SHOPIFY'S INTERNSHIP APPLICATIONS ARE OPEN! .- Canada Winter 2026 (.- USA Summer 2026 (. Work on full time teams, ship to millions of merchants, using the _right_ models (no token budgets here😏). Apps close Sep15 @ midnight ET.
2
0
23