Vishal Dharmadhikari

MTS, @GoogleDeepMind
@GoogleAIStudio, Gemini API
My views and opinions

[Vishal Dharmadhikari @Vish_owl]

If you love working at the frontier and building things, come join us!

[Omar Sanseviero @osanseviero]

We're hiring our first FDEs!

Being a FDE at Google DeepMind has a fun flavor. You will:
🤝 Act as a technical bridge between research teams & strategic partners
🛠️ Build benchmarks and evaluation tooling
🚀 Work at the frontier of AI

Join us 👉
https://t.co/3ybAhNZngz

[Vishal Dharmadhikari @Vish_owl]

Great guide on using Managed Agents w/ Weave tracing!

[Weights & Biases @wandb]

The scariest bug from a coding agent isn't the one that crashes.

It's the one that runs cleanly, passes tests, and quietly produces wrong results.

So we built one on @Antigravity's Gemini Managed Agents API to hunt them.

Give it a repo. Get back the bugs that passed review.

[Vishal Dharmadhikari @Vish_owl]

@OfficialLoganK More pie chefs needed

[Vishal Dharmadhikari @Vish_owl]

@tekbog @basedjensen @OfficialLoganK @zapier Email sent!

[Vishal Dharmadhikari @Vish_owl]

@basedjensen @OfficialLoganK @zapier @tekbog Happy to help support on this! Let me know if there's anything we can do to help

[Vishal Dharmadhikari @Vish_owl]

@theXipuLi Would you be able to shoot me a DM with more details? Happy to help debug this with you!

[Vishal Dharmadhikari @Vish_owl]

@kkpattern @OfficialLoganK Hey Kai, you just need to return thought signatures

[Vishal Dharmadhikari @Vish_owl]

@Eigent_AI 🔥 demo!

[Vishal Dharmadhikari @Vish_owl]

@rene_sultan @GoogleDeepMind 🔥 it was awesome partnering with you on this!

[Vishal Dharmadhikari @Vish_owl]

@samsheffer Pokemon was such a formative childhood experience

[Logan Kilpatrick @OfficialLoganK]

prompt to profitable company, this is the future

[Vishal Dharmadhikari @Vish_owl]

No better workhorse models than the Flash models!

[News from Google @NewsFromGoogle]

New stat from @vercel's AI Gateway in @BusinessInsider: Gemini 3 Flash is leading across AI models in token usage as of April. 🚀 See more stats on how developers are using our models → https://t.co/0xbulg4X2S

📊via @BusinessInsider

[Vishal Dharmadhikari @Vish_owl]

@matthewridenour @GoogleDeepMind @calm Love it! One of my favorite case studies

[Vishal Dharmadhikari @Vish_owl]

@LukeParkerDev How do I sign up

[Vishal Dharmadhikari @Vish_owl]

Can you set up an autonomous fine tuning loop entirely from your phone?

As a fun experiment, I used my @NousResearch Hermes agent to orchestrate @karpathy's autoresearch to improve @googlegemma 4 E2B's tool calling perf on BFCL. @modal made it v easy to spin up the necessary compute and run the experiments.

All of this set up, all from my phone! It took a bit of debugging, but the pipeline is finally stable and running experiments.

I also used the Gemini API's new flex inference to cut costs on the scientist by 50%. Since this is a long running task, the cost for latency trade off was worth it.

Regardless of end performance, being able to do this all from my phone is a pretty cool glimpse into the future of agentic UX!

[Vishal Dharmadhikari @Vish_owl]

@bernaferrari @reah_ai If you're using AI Studio, you can enable Logs for your project. This allows you to track all the requests made via the project

https://t.co/2uEkwwnqj0

[Vishal Dharmadhikari @Vish_owl]

@mathlover3b1 @rekram11 @LukeParkerDev Would you be able to send me a DM? Wasn't able to send you one.

[Vishal Dharmadhikari @Vish_owl]

@rekram11 @LukeParkerDev Sent you a DM!

[Vishal Dharmadhikari @Vish_owl]

@MaximeRivest Sorry about that - we're working on stabilizing things