[Baypiggies] BayPiggies Jan. 23, 2025 - Packages, Peaks, Massive Graphs, Resilient Execution

[Karen Dalton]

*BayPiggies Jan. 23, 2025 meeting!

Packages, Peaks, Massive Graphs, Resilient Execution*

  * *https://www.meetup.com/baypiggies/events/305328742 (in-person at
    SAP Labs in Palo Alto)*

  * IMPORTANT: SAP requires all registrations be completed /in advance/
    - and will email you prior to the event -- so no last minute
    registrations will be honored, and no walk-ins will be allowed.

    **Please RSVP on Meetup and also fill out the Google form here /*by
    January 20*/ so you can receive the SAP form before the event
    (REQUIRED): https://forms.gle/pCkGsCrzr3txS2VS6**


*Schedule:*
- 6:30 Register and Refreshments
- 7:00 Welcome + Announcements
- 7:05 Lightning Talk - Popular Python Packages Series - "uv"
- 7:10 Short Talk - Peaks, Valleys, and Python: Pipelining Biochemical 
Data Processing with hplc-py
- 7:30 Short Talk - Cosmograph for Python — a new library for 
visualizing massive network graphs and machine learning embeddings
- 7:55 Main Talk - DBOS Transact: Ultra-Lightweight Durable Execution 
for Python
- 8:30 Wrap-up

***** Lightning talk ******

*Popular Python Packages Series - "uv"*

Karen starts off a new series of short talks in 2025 about trending 
Python tools and packages, starting with "uv", an extremely fast Python 
package and project manager.


***** Short talks ******

*"Peaks, Valleys, and Python: Pipelining Biochemical Data Processing 
with hplc-py" presented by Griffin Chure
*
A common task in biological data processing is taking a complex signal 
and breaking it down into its constituent parts. However, biological 
data is noisy and evaluating the accuracy of an analysis pipeline often 
requires manual intervention and assessment, hampering the throughput 
that scientific or industrial problems often demand.

In this talk, I will highlight `hplc-py`, an open-source Python library 
I developed to tackle this problem in the context of chromatography — an 
analytical technique for quantifying the components of chemical 
mixtures. This package leverages diverse functionality in the 
scipy-stack — from peak detection to parameter optimization — to 
deconvolve complex chemical spectra into signals from individual 
molecules. While state-of-the-art software addressing this problem 
requires extensive manual intervention, `hplc-py` significantly 
decreases this intervention and is designed to be easily integrated into 
data analysis pipelines. I will highlight how this software was 
developed to be robust across different experiments while allowing the 
end users to rapidly interact with the analysis and validate the results.

While the processing of real scientific data will be demonstrated in the 
talk, no scientific background will be assumed. The open-source library 
is available on GitHub (github.com/cremerlab/hplc-py) with detailed 
documentation (cremerlab.github.io/hplc-py).

Griffin is a computational biologist with broad experience leveraging 
mathematical modeling, Bayesian statistical inference, and scientific 
software engineering to decode the complex mechanisms governing cellular 
function and behavior. He is passionate about building performant and 
robust software that employs quantitative methods to dissect biological 
data, and building strong collaborations with scientists and engineers 
across disciplines.

*"Cosmograph for Python — a new library for visualizing massive network 
graphs and machine learning embeddings" presented by Nikita Rokotyan
*
Visualization of large amounts of data has always been a struggle and 
required having sophisticated workflows. Cosmograph, a JavaScript 
library for creating stunning, interactive data visualizations, is now 
accessible to Pythonistas, data scientists, and AI engineers from the 
comfort of their notebooks.

Cosmograph can handle several millions of nodes with GPU-accelerated, 
force-directed layouts, enabling real-time exploration of 
multidimensional data and complex networks. It’s the fastest JavaScript 
library for Network Graph visualizations and its interactive tools—like 
zooming, panning, filtering—help to transform data chaos into clarity.

This talk introduces the audience to Cosmograph’s capabilities and 
demonstrates how it integrates seamlessly into Python workflows. No 
specialized expertise is required, though those involved in data, 
machine learning, or AI are likely to find it particularly engaging.

Cosmograph is available on PyPi (https://pypi.org/project/cosmograph/) 
with a documentation at 
https://cosmograph.app/docs/cosmograph/Cosmograph%20Python/get-started-widget.

Nikita (https://rokotyan.com) is the founder of Interacta, an 
award-winning team of scientists, designers, and engineers dedicated to 
building beautiful and functional tools for data exploration. Beyond his 
passion for data, he has a deep appreciation for new media art and music 
and enjoys discovering the stunning landscapes of California.


***** Main talk ******

*"DBOS Transact: Ultra-Lightweight Durable Execution for Python" 
presented by Peter Kraft
*
We present DBOS Transact, an open source Python library providing 
ultra-lightweight durable execution. Durable execution means your 
software is resilient to failures. If ever interrupted or crashed, a 
DBOS Transact application can resume from the last completed step, 
automatically.

Under the hood DBOS Transact works by storing your program execution 
state in a Postgres database. There's no need for an "orchestration 
server": all you need is a Postgres database and adding some lightweight 
Python decorators to your code. This approach is incredibly 
cost-efficient and performant.

We will present some cool features of the framework such as scheduled 
jobs and exactly-once events processing.

Peter is a co-founder of DBOS, Inc., building a new serverless platform 
for backend developers that radically simplifies backend development. He 
co-founded DBOS based on his PhD work at Stanford, where he was advised 
by Peter Bailis and Matei Zaharia and worked closely with Michael 
Stonebraker. He is interested in databases and distributed systems.

Important attendance note: SAP requires everyone to sign a 
confidentiality and security disclosure to maintain confidentiality and 
adhere to SAP's physical security protocols during your visit to SAP's 
facility. This applies to all guests. All attendees must show a 
Government-issued ID, and sign the SAP Security form to enter the event.

Register by Monday (midnight) January 20!

*/Thank you, SAP Labs, for sponsoring and hosting this month's meeting!/*

----

*Personal Donations to BayPiggies via PSF: *Please consider supporting 
future BayPiggies events and Python in the Bay Area at the link below 
via the Bay Area Python Association and the Python Software Foundation:

https://psfmember.org/civicrm/contribute/transact/?reset=1&id=43


----


See you soon!
-Karen and the BayPiggies organizers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/baypiggies/attachments/20250107/925ef502/attachment.html>