Tip Sheet #77: Upcoming events with free virtual pass
Hi Tip-Sheeters,
This week I have a few odds and ends to share. There are a few events coming up (including one with a free virtual pass) and a couple of newer ML models you can explore.
Upcoming events in tech and data science
I am a fan of vendor-neutral events in the tech space. The speakers at these events have a bit more freedom to share their experience without promoting a product or company. Here are a few worth checking out:
Open Data Science Conference West 2026 - Oct 27-29
The strength of this conference is that it gets really good speakers like Diogo Almeida of Jev fame (read on for more on that topic). It’s both in-person in San Francisco, CA, and virtual.
Best of all: here’s a free pass for the virtual portion. Highly recommended.
APIDays Paris - Dec 1-3
I’ve worked with APIDays for several events and courses, and they really do a great job. The Paris event is their largest event and is combined under the same roof with a lot of other events on AI, green computing and a variety of other topics. The location is in La Défense in Paris’s financial district.
PyDay Lima 2026 - Dec. 11
This is a one-day event in Lima, Peru. (In-person only.) The theme this year covers data science, robotics, and automation, among other topics. Looks like they’re still open for speakers if the location works for you.
New ML Models to Check Out
While LLMs receive most of the oxygen these days, other ML models have advanced too. Here are a few special-purpose ones worth exploring:
Jev from TypeSafe
Jev has a different job than LLMs. Instead of getting text and returning text, Jev is a System One Model.
For one practical example explained by unicodedeveloper, you can send a help desk ticket to Jev and get a decision for the best department to handle it. It's much faster and cheaper than using an LLM for this task.
Tip-Sheeters will be very familiar with the deployment model of Jev: It’s a web-hosted API. You can use it directly by signing up for an API key through their developer portal. Or you can use it through an AI gateway like Vercel's. Both methods require payment -- inference ain't free.
TimesFM-3 from Google Research
Time series forecasting is a core data science task. You have historical data that’s tied to dates and times, and you want to predict outputs in the future. One method of accomplishing this task is training a model yourself using your historical data.
Using a different approach, TimesFM is a zero-shot model built for time forecasting. It can forecast from your historical data without being trained on it or needing fine-tuning.
You can find the PyPI package here for non-commercial usage: https://pypi.org/project/timesfm/
That's it from me, have a great week!
Keep coding,
Ryan Day
👉 https://tips.handsonapibook.com/ -- no spam, just a short email every week or so.
