Ep59: NODES 2025 Highlights + Solving Graph Problems with Cypher 25

In this episode:

• Recap of NODES 2025 and standout sessions
  • How AI and music graphs are shaping new tech (featuring Luanne Misquitta’s talk)
  • Exploring RushDB: open source tools for graph data
• Developer advocacy in the classroom: inspiring the next generation
• Updates on Spring AI, Langchain4j, and upcoming workshops
• Blog post on new Aura Fundamentals course
• Solving tough graph problems with Cypher 25

Resources Mentioned:

• NODES 2025 playlist (only keynotes at this time)
• Luanne Misquitta’s Music Graph session
• RushDB session by Artemiy Vereshchinskiy
• Langchain4j read only db issue (solved!)
• Neo4j Graph Academy Aura Fundamentals blog post
• Solve Hard Problems with Cypher 25 blog post
• Advent of Code (2025)

Thanks for listening!

Episode Transcript

(00:05):
You are listening to the BreaktimeTech Talks podcast, a bite-sized tech
podcast for busy developers where we'llbriefly cover technical topics, new
snippets, and more in short time blocks.
I'm your host, Jennifer Reif, anavid developer and problem solver
with special interest in data,learning, and all things technology.
By the time this is published, NODES2025 will be over and many of the

(00:29):
attendees, staff, and speakers willprobably still be feeling the after
effects of the adrenaline rush.
This time of the year is also somewhattricky because I am wrapping up
commitments and goals for 2025, whilealso still preparing for content and goals
for 2026, it's exciting and very busy.
It is also probably my favorite timeof the year because I can accomplish

(00:50):
a few final things for the year,take a bit of a breath, and prepare
for exciting adventures that arecoming in the next year, 2026.
I wanted to highlight a coupleof things from NODES 2025 today.
The first one is that I heard a greatpresentation from Luanne Misquitta
about graphs and AI on a music graph.

(01:11):
If you're not familiar with Luanne'swork, she's done a ton of content and has
been in the Neo4j community a long time.
She's written a lot of blog posts,written books, and been featured
at NODES events and Neo4j events.
I always really enjoy hearing herstuff and reading her content, but
this one really spoke to me, this year.

(01:32):
First of all, it's on a music graph,so as a fellow music nerd, I definitely
plan to dig into this a bit more.
And apparently this project is alsothe focus of her latest book, so I
hope to read that here soon as welland also review the video that will
be released on YouTube of this notes2025 session here in a couple of weeks.
This particular presentation, though,looked at piecing together similar

(01:57):
tastes in music or recommend a playlistprogression from one style to another.
An extra bonus is that this applicationused Spring AI, which of course you
know is near and dear to my heart.
I definitely am excited to see alittle bit more about this project,
understand how it was constructed,and the cool things that you can

(02:18):
do with AI, music and graphs.
Another session that stuckout to me was on RushDB.
RushDB is an open source toolthat uses Neo4j under the hood and
offers a really nice interaction andimport strategy and capabilities.
I thought the presenter did a reallygreat job showcasing the product and

(02:40):
speaking Everything was very casualand comfortable, and I just really
enjoyed listening to this sessionand am really excited to learn more
about Rush DB and try some things out.
Now, I will say there were tons ofsessions that had varying degrees
of speaking expertise, stages ofproject, and different types of
companies and different industries.

(03:01):
It was really cool to see all thedifferent things that people are
doing with graphs, and I must sayeven those who were new to the space,
new to graphs or new to speakingeven, I really admired their bravery.
NODES is yes, a community event, butthere are thousands of people that
get online, watch the videos or attendlive, and for folks who are new to

(03:23):
speaking and new to graphs, it took anincredible amount of bravery to speak
and talk about their projects and whatthey are doing in the technology space.
If you missed it, don't worry.
All the recordings will be live onYouTube here in a couple of weeks.
And actually the keynotes are alreadyavailable on the YouTube playlist.
I will leave a link to those in the shownotes, as well as links to the sessions

(03:46):
that I called out here earlier fromLuanne Misquitta and the one on Rush db.
Also, to add another layer ofexcitement to this week, I spoke
to a local eighth grade careerclass about developer advocacy.
First of all, I lovedoing this type of thing.
It's always fun to me to talk tostudents and young people and encourage

(04:09):
them in whatever industry or careerthey might be looking into, showing
them that there are lots of thingsyou can do outside traditional job
role descriptions, if you will.
The class was really attentive andthey asked a lot of great questions.
Of course, they zeroed in on theworld travel because that tends to
be the shiny part of my job role.

(04:30):
But they also asked about my day-to-dayactivities and how I balanced both
personal and professional lives.
I was really excited to share theadventures beyond traditional engineering
roles, as well as my unique pathto reach my career in technology.
I mentioned a few weeks ago that I wasworking on a workshop project, and I

(04:50):
built an application with Spring AI,but I wanted to create a Langchain4j
version of that application But I couldn'tbecause I was using a read only database.
At the time, Langchain4j required youto have write access to the database
in order to create the vector indexso that it could make sure it was
there and do the vector search.
At the time, it wasn't possible for meto use the read only database that I

(05:12):
was using and have a Langchain4j app.
That issue is officially solved.
I created a GitHub issue a coupleweeks back when I had the problem
at the first place, and someonecreated a PR and that got merged.
I have yet to try this out, but I didreview the PR changes and those look
solid, so I'm really excited to test thisout and build the Langchain4j version

(05:34):
of the Spring AI app I already have.
I will keep you posted on theupdates to that and how that goes.
I also will have some more updatescoming soon on the various projects
that I'm currently working on.
I do have a couple of upcomingworkshops that will be in December.
I'll keep you posted on thetopics and where you can join
those, if you are interested.
Then I also have 2026 conferences linedup for pretty much all winter and spring.

(05:58):
It's kind of odd and surreal to methat I already have pretty much the
first half of 2026 lined up with myevents already, but I'll talk through
some of those as time moves along.
Then I also worked onthe Java book this week.
I did a lot of focus on trying to getsome things built out there, get some
things accomplished in that project.
And I'll also talk about in the future,the applications I'm building and the

(06:22):
topics that I'm exploring, either theend of this year, early into next year.
The content pieces I wannahighlight, first of all, NODES
2025 playlist, obviously.
I will leave that linked.
Again, the only thing that's sittingout there right now are the keynotes,
but definitely take a listen to those.
There were some really fabulous contentand super exciting insights there.
I watched them at least once and thenI went back and watched pieces of

(06:44):
them again, and I definitely will bedoing that yet again in the future.
And then, of course, the rest of thesessions will be available on that
playlist as well here in the coming weeks.
Then I also read a couple ofreally interesting blog posts
on the Neo4j developer blog.
The first one is that there is a new GraphAcademy course for Aura Fundamentals.
My colleague put a short blogpost out about this course.

(07:08):
GraphAcademy has free self-paced onlinecourses that you can work through.
This one talks about Aura,specifically our database as
a service product at Neo4j.
It talks about the different tiers,the features, the tools, and et cetera.
The blog post is short.
It just gives the high levelhighlights of what the course will
cover, and then of course gives youa link to check out more information

(07:30):
and to do that course on your own.
The other blog post that I read was "SolveHard Graph Problems with Cypher 25".
Our engineering department, the lastseveral months, has been putting out some
really great content, specifically blogposts, about how to use the new features
in Cypher 25 or things that you couldn'tdo before that now you can in Cypher 25.

(07:53):
This has been really exciting to me.
I always like to learn more aboutCypher, how I can get better at
structuring my queries, and thisparticular blog post did not disappoint.
It pulls up a challenge fromthe advent of code in 2021.
If you're not familiar with theadvent of code, it's daily challenges
throughout all the month ofDecember that stretch your coding

(08:14):
capabilities and are just fun puzzles.
If you wanna learn more about that,I will link the advent of code
website in the show notes as well.
But this one pulled a particular, day12 challenge from the year of 2021.
It's where they modeled submarinecave systems as a graph.
they took a submarine cave system,pushed that into Neo4j and modeled that

(08:37):
in the graph database, which was cool.
I've never seen this use case before.
I find this really fascinating.
The author walks through the blogpost, though, that the solution to
solve the problem in 2021, felt alittle bit more like a workaround.
He knew there was a solution whenyou went seven hops deep in the
graph, so he was able to piece thattogether and construct the result.

(09:00):
But again, that felt morelike a. workaround rather than
a natural analysis result.
The difference between what was possiblethen and now is that it went from a
different relationship requirement.
Cypher used to require that youhad a different relationship,
which ensured that you would onlytraverse a particular path once.

(09:22):
Now, with Cypher 25, you can haverepeatable elements, which allows
you to traverse certain paths orrelationships a specified number of times.
Now, this larger search space couldresult in path explosion, but there
is a function to help with that calledallReduce, which is applied during
the traversal and ex ensures thatthe performance remains really high.

(09:45):
The allReduce function will track thesmall cave visits during the traversal
by building a small visited list, thecondition then ensures that exactly
one small cave can be visited twiceand prunes the paths that violate
this rule during the traversal.
This is a massive improvementfrom the previous example in 2021,

(10:05):
where you had to do post filtering.
You would suck in all of the pathsand traverse all of the paths.
And then post filter, thepaths that didn't meet that
requirement, that criterion.
As you might imagine, pull inall those resources, drain the
memory, then prune post filtering.
The allReduce avoids that by keepingthe possible pads to a minimum,
dropping those unnecessary pads that youtraverse as you move along in the query.

(10:30):
It's been an exciting week.
NODES 2025 was packed with great sessions,and I'm still thinking about it all.
On the project side, I'm happy that theLangchain4j read only database issue
that I mentioned before is updated.
I'm looking forward totesting it out soon.
I'll be back with updates onthe new workshops and what's
also coming up for 2026.
In the meantime, t hanks forlistening and happy coding.