Franziska Hinkelmann, PhD
@fhinkel
Engineering Manager @Google Cloud. @nodejs TSC member. Former @v8js. Puppies, tacos, and LLMs.
Interested in learning a bit about V8's bytecode? medium.com/dailyjs/unders…
Can't wait for OAuth!
Results are in! 🧩Prompts > 🔐MCP OAuth > 📚Resources > 🧠Sampling Here are some links if you're unfamiliar. - Prompts: modelcontextprotocol.io/docs/concepts/… - MCP OAuth: modelcontextprotocol.io/specification/… - Resources: modelcontextprotocol.io/docs/concepts/… - Sampling: modelcontextprotocol.io/docs/concepts/… Time to cook!
Any Reddit friends who can help here?
So supposedly @Reddit is marking all of my comments as spam despite me being a moderator. Any action to "approve" results in errors. Anyone have any advice on how to resolve this? Happened in the Gemini CLI AMA as well...
Gemini 2.5 Flash-Lite is now GA, signifying it is ready for production use! It's the fastest and lowest cost ($0.10 input per 1M, $0.40 output per 1M) model in the Gemini 2.5 model family. I tested it with Agent Development Kit (ADK) and it's speedy! (video is in real-time).
Gemini 2.5 Flash-Lite is now available: it's our most cost-efficient and fastest 2.5 model yet, ideal for low latency and high-volume use cases. developers.googleblog.com/en/gemini-25-f…
Google DeepMind is winning, they just don’t hype it. > Participated in IMO > IMO officials verified the solution > Awarded Gold medal > Quietly dropped the post on a Monday morning At this point, they’re not even bothered about competing. Also this version of Gemini is coming…
Life pro tip: don't pick up sticks covered in (fire?) ants for your dog.
I sat down with the #AllThingsOpen team to chat about AI allthingsopen.org/articles/getti…
I love this!
Met with a googler today who found Gemini CLI and started using it at Google and he couldn't contain his excitement: The TLDR for my past week is I went from "I don't get why everyone is so excited about a CLI for Gemini" to "I think this changes how I think about engineering"
ohhhhh, that's a pretty color theme 😍
For production use, you'll want to interact with your model securely. You can use the Google Gen AI SDK and pass an identity token with your requests to ensure authenticated access. 🔐 Here's a quick example in JavaScript/TypeScript. #Security #Auth #SDK #JavaScript
Great read! Cc @karl
Implementing High-Performance LLM Serving on GKE: An Inference Gateway Walkthrough cloud.google.com/blog/topics/de… < Some of the best Kubernetes work (in open source, or in commercial products) still happens at Google. Great post from @kweinmeister
On the plane, vibing to my YouTube playlist🎧. Would be vibe coding, but the flight attendant asked us to put our devices away.
Help us define the Gemini CLI roadmap! v1 is coming. Feedback welcome github.com/google-gemini/…
Tired of single-agent AI limitations? The future is in multi-agent systems. 🤖 But how do you build them without the chaos? This video breaks down the essential architectural patterns (like Supervisor & Sequential Flow) you need to know. A must-watch for devs! #AI #LLMs…
Great opportunity!
I’m looking for few people who are either experts in TS or Python to work closely with me on a few next generation AI developer frameworks and tools at Google. If you are interested in this overlap, my DMs are open for resumes!
GEMINI.md
if only we had a place to put all this data instead of random text files. like a base. like a base for data ...
My life saver!
Gemini CLI can automatically remember things for you. Need to take control? Use the /memory slash command. Memories are stored in your ~/.gemini/GEMINI.md
Some interesting Gemini CLI use cases and tutorials 🧵⬇️
Less than 149 hours ago, Google dropped Gemini CLI. Gemini 2.5 Pro, high usage limits, FREE. Minds are blown. And people are already coming up with wild use cases. 10 examples: