home.social
  1. Finally, *Senior Professional Officer: DevOps Engineer* (*Senior Engineer*)

    Like many data teams, we own our specialised data infrastructure. This includes @airflow, and (along with for querying), as well as interactive platforms such as @ProjectJupyter and @posit_glimpse. We run a cloud native, open source stack, with @k3sio for our on-prem setup, and have a growing cloud presence.

    Come help keep the magic happening!

    Apply at www.capetown.gov.za/careers

    6/6

  2. Next, *Head: Product Services*

    We've been heavily influenced by people like @richardjpope, @pahlkadot, the , etc. - not having is why a lot of doesn't work. We're fixing that by hiring the right people.

    We understand good product management to be about making sure that we're building what is needed. This is hard with , where there is so much hype and misunderstanding! Come help us do better

    Apply at www.capetown.gov.za/careers

    3/6

  3. *The purpose of a system is what it does*

    Please share with anyone that needs this, or might know those that do

  4. Are you a interested in getting into , but want to still keep a toe-hold in acdaemia?

    Our frequent collaborators at Africa have a post for a data engineer that will embedded in the City of 's Data Science team: povertyactionlab.org/careers/d

    Feel free to reach out with any questions

    (please boost for reach)

  5. Need to do some load testing of an internal HTTP service, mostly emulating interactive workloads. Primarily interested in measuring latency and throughput. Any recommendations?

  6. Not quite the meaning of delivery I had in mind while setting up this news alert....

  7. Doing a talk about APIs (and really, service architectures) for our City of Data Community of Practice - my audience is smart, and somewhat technically literate. What's your favourite way to explain modern service architectures?

    Looking at plenty of @b0rk comics for inspiration!

  8. Well, it seems most of is without power. Is giving me a good opportunity to appreciate the new UI of @eskomsepush!

  9. CW: PyConZA thread 12/n

    Annnddddd finally, @czue finished things off in grand style, talking about his experiences of generating passive income. It served as a really great 101!

    The Tao seemed to be about building simple, effective apps, with a focus on quick iterations, bootstrapping, and testing your product early. Something I really liked was how the real goal of indie hacking isn't necessarily great wealth, but freedom and creative agency! 💻🌱

  10. CW: PyConZA thread 11/n

    I was then busy as the organisers ill-advisedly gave me a slot, talking about *You don't need a data service, you just need an object store and some JSON files*.
    Basically, you can eschew a lot of the complexity of building a service behind an API, if you can actually just get away serving up files from an object store like , or . Putting a CDN in front then makes it really scaleable and cheap!

    Slides here: talks.cct-datascience.xyz/pyco

  11. CW: PyConZA thread 9/n

    Next up, Johan Beyers (in his own words, *has seen some stuff*) shared how use @ApacheKafka for Change Data Capture, Event Sourcing and Task Queues. Definitely gave me a lot to think about, in terms of how data flows across distributed systems. He also had lots of great content on designing components like messages, topics, producers and consumers.

    Sounds like a good choice for handling large-scale, high-throughput environments! 🚀

  12. Had an excellent two days at in Woodstock, .

    Yesterday we launched Africa's first City , and today we had a great workshop led by some excellent experts from orgs such as , , and

    Check it out - resource.capetown.gov.za/docum

    Also, it's just nice to be in a room with my colleagues.

  13. A workflow I tried out, with reasonable success, was using my shiny new Kindle Scribe to take notes at , then used our Llama 3.1 model to clean up the inevitably mangled OCR output, and organise my thoughts into a series of pithy posts.

    Results in this thread: elk.zone/fosstodon.org/@Gord1i

    The results weren't magic, as I still had to proof read, and adjust to my tone, etc. but definitely got me a good 60% of the way there.

  14. CW: Dataconf 2024 Thread

    Finally, @gbelrose continued his theme from last year of applying good software engineering to data engineering, and now reporting. This was pretty inspiring, as it's very tempting to throw your hands up when you encounter a mess of dashboards created via elaborate clicky-clicky.

    G advocated using platform APIs to extend your data pipelines into this domain, introspecting and annotate dashboards.

  15. CW: Dataconf 2024 Thread

    Another great talk was from Zintle Faltein-Maqubela on her experiences running data-focused internship and graduate programmes. Was well timed, as we're hopefully going to be taking on more juniors soonish!

    The big point I got was around managing expectations on both sides - particularly for interns, they've often got other things going on in their lives. So getting some understanding of that, against what their team needs from them.

  16. CW: Dataconf 2024 Thread

    Also, really enjoyed Gerhard van Deventer from on *Lessons Learned from Using an LLM and Expecting Magic*.

    Solid coverage of fundamentals that really matched our experience - Improve your prompting; eval early, eval often; break big tasks into small ones -> set up a flywheel, and start eeking out those iterative improvements.

  17. CW: Dataconf 2024 Thread

    Over and above James's cool use of interactive polls in his keynote, there were many great points about how to be part of the "community". Something that was key was recognising what you take vs what you give back varies over your career.

    And of course, he ended up with a rousing rendition of You Must Check Accessibility (YMCA).

  18. CW: Dataconf 2024 Thread

    First up, keynote from James McGillinray!

    Talking about the value of community in the context of crazy, exponential technology change. A topic close to my heart

  19. Looking forward to tomorrow in not so sunny ! Looks to be some great talks in the line up.

  20. Have a PR out for the spec for the data API that backs @coct_service_alerts - github.com/cityofcapetown/serv

    The intent here is to capture the current situation, before following on with a new PR for the `v2` API paths.

    Comments welcome!

  21. Recently listened through 's in pretty much one sitting while driving cross country.

    As someone with a passing interest in cryptocurrencies, effective altruism and probabilistic thinking, it was pretty crazy to hear what happens when it's all turned up to 11.

    A bit conflicted about how mental health is mentioned but not directly addressed in the book. It was probably responsible to not armchair diagnose, but it was a big piece of the puzzle to leave out.

  22. CW: Dataconf thread

    Really enjoyed Guillaume Belrose's talk on TDD for data engineering. Will definitely be checking out his unit testing package.

  23. CW: Dataconf thread

    Next up - Johanna Thathaisa on Low-Code Data Tools!

  24. CW: Dataconf thread

    Actually, is turning into an enjoying take down of the large consulting world.

    Strategy as a service model:

    1. Make overblown promises, driving ill-advised adoption of tech (e.g. big data, AI, LLMs)
    2. Learn from other people's inevitable failures
    3. Sell superior version of (1)

  25. Great to be at in

    First up is Jade Abbot on why you need to know where your data is!

  26. @danie10 Ja, have done excellent work there, as well as with Vulekamali, National Treasury's more general portal (vulekamali.gov.za/) - JD did a great talk on building it at last year's PyConZA - youtube.com/live/JSfg-H5XIHI?s

    have also done some good stuff, particularly with narrative content on the Durban Edge portal - edge.durban/

    These are all for dissemination though - I would like to push the envelope, along the line of O'Reilly's gov-as-a-platform idea.

  27. Getting into @pahlkadot's Recoding America (which is amazing) and it's eerie how relevant it is to my experience in local gov on the other side of the world.

    > Elites understand policy. They are comfortable with it. They may also be comfortable with politics, which is often described as "how the policy gets done." The rest of us know about delivery.