Felipe Hoffa

It’s exactly 10 years since my first day at Google. What I’ve learnt: 1. In 2011 there was a career path for me, but I had to leave Chile to find it. (thread continues)

I compared all these models with @yupp_ai. Find the full responses on yupp.ai/share/47703833… <-

My new hobby: Figuring out the best LLM to make charts and drawings with SVG. In this case I asked the LLMs to "draw an svg chart of the top 10 major cities in North America by perfect days, include a happy sun", and you can see here the results. linkedin.com/feed/update/ur… <-

Follow the rest of the discussion on linkedin.com/posts/hoffa_7-… (I'm way more active there than here)

For proof, here's me asking 10 LLMs through @yupp_ai: yupp.ai/chat/7578801d-… - they make it really easy to do your own evals.

7 out of 10 LLMs got this simple geographic question wrong. Can we trust AI with even basic questions? The question and the correct answer: youtube.com/watch?v=f33-IR… The shocking part: Google @GeminiApp, @MistralAI, @OpenAI o3... all got it wrong. Trust - but verify.

How is data made? On Saturday I boarded a 7am bus to Reno to figure it out. From algorithms to door-knocking, this weekend changed how I see campaign data - and made me a better data professional too. Here's what I learned: linkedin.com/posts/hoffa_ho… 🇺🇸

@felipehoffa @cbergman so great to see you at the @awscloud GenAI Loft today!

I’m ready to announce my newest “label”. I recently reflected on how I’m taking some time-out to figure out who “Felipe” is, leaving behind the “from @SnowflakeDB" and “from @Google” labels. Well, I went yesterday to a life changing interview, and... linkedin.com/feed/update/ur… 🇺🇸

If I lived anywhere else in the world, I'd visit San Francisco. There's so many opportunities to learn, discover, and meet others in the tech world. And I even won the 3rd place at an @awscloud AI hackathon! More details: linkedin.com/posts/hoffa_if… ✈️🌁

@benthompson Hard agree, but speaking of embracing change, I'd rather read you on @Meta Threads instead.

Every tech company *cough* Meta *cough* needs to accept that YouTube won and stream every event there. Put it on your site if you want, but put it on YouTube too.

❤️ twitter.com/civitaspo/stat…

❤x.com/rakudeji/statu…

All good things must come to an end: After 3.9 years at @SnowflakeDB, it’s time for me to say goodbye and thanks: - linkedin.com/feed/update/ur… ❄👋

@kurtschwehr @GlobalFishWatch Thank you Kurt! It would be cool to collaborate on that kind of project again :)

Felipe helped us get started on BigQuery for the All the Ships in the World demo that led into @GlobalFishWatch. Looking forward to seeing what he is up to next.

@PetrSimecek Yeah, I'm sure other people are leaving too. But I'll let them tell their story.

@felipehoffa Nobody else from your Snowflake network?

Hi @felipehoffa! I need your help. A regular user can load data from AWS S3 using the COPY INTO command, and the same applies to Azure. However, in Google Cloud, a user must create a storage integration, which requires either ACCOUNTADMIN access or a user with global CREATE INTEGRATION privileges. Why is this the case? Why can a user like 'John' load data from his S3 into his designated schema, but when it comes to GCS, he has to create a JIRA ticket and wait patiently for weeks until someone from his IT service desk processes his request to create a storage integration? John typically doesn’t have the rights to create a storage integration, and he can't provide GCS credentials to his GCS stage in Snowflake. What is the purpose of creating an additional obstacle for users utilizing GCS compared to those using S3 or ABS? As a side question: Is there any way for a regular user without extensive privileges to fetch data from GCS?