MLOps in Jenkins-X
As discussed in the SIG meeting, here is a link to a quick demo
video of MLOps within Jenkins-X. Things are developing fairly
quickly and we plan to have a broad range of template projects in
our public repo in the new year.
https://youtu.be/AqL_ME7BM6U
As discussed in the SIG meeting, here is a link to a quick demo video of MLOps within Jenkins-X. Things are developing fairly quickly and we plan to have a broad range of template projects in our public repo in the new year.
https://youtu.be/AqL_ME7BM6UTerry CoxBootstrap Ltd
On 7 Jan 2020, at 01:12, Michael Neale <mneale@...> wrote:
That is pretty cool Terry.With the training repo - is the idea that datascience type persons work in that, play with their data/model/notebooks, and when ready commit or issue a pull request to the *training* repo? do they then look at the PR in the service repo to test the real thing out? what is the workflow between data science experts and the engineers that work on the service/app that you see out there?On Fri, Dec 20, 2019 at 5:41 AM Terry Cox <terry@...> wrote:As discussed in the SIG meeting, here is a link to a quick demo video of MLOps within Jenkins-X. Things are developing fairly quickly and we plan to have a broad range of template projects in our public repo in the new year.
https://youtu.be/AqL_ME7BM6UTerry CoxBootstrap Ltd
--Regards,Michael Nealetwitter: @michaelneale, skype: michael_d_nealeCell: +61 423175597 (Australia)Cofounder @ CloudBees
The intention is to get the Data Scientists out of the habit of working in isolation on their laptops for months and focus them on being regular and iterative collaborators to an MLOps team in a distributed production-capable environment.In practice, most of the DS team seem keen to maintain the service component of their ML solution and as a result are learning very rapidly about the practical challenges of scaling distributed solutions effectively.Terry CoxBootstrap Ltd+44 (7797) 828891+44 (845) 287-2345On 7 Jan 2020, at 01:12, Michael Neale <mneale@...> wrote:That is pretty cool Terry.With the training repo - is the idea that datascience type persons work in that, play with their data/model/notebooks, and when ready commit or issue a pull request to the *training* repo? do they then look at the PR in the service repo to test the real thing out? what is the workflow between data science experts and the engineers that work on the service/app that you see out there?On Fri, Dec 20, 2019 at 5:41 AM Terry Cox <terry@...> wrote:As discussed in the SIG meeting, here is a link to a quick demo video of MLOps within Jenkins-X. Things are developing fairly quickly and we plan to have a broad range of template projects in our public repo in the new year.
https://youtu.be/AqL_ME7BM6UTerry CoxBootstrap Ltd
--Regards,Michael Nealetwitter: @michaelneale, skype: michael_d_nealeCell: +61 423175597 (Australia)Cofounder @ CloudBees
On 8 Jan 2020, at 00:20, Michael Neale <mneale@...> wrote:
Right - so they would contribute to both repos in this case? Does that mean there are developers embedded with the DS team? (or they take on the role of developers - similar to devops share responsibilities with ops?)Really interesting... I guess the benefit of quickstarts and opinionation is they don't have to discover this stuff, but still get some rigor/process in place. Interesting to watch this evolve.On Wed, Jan 8, 2020 at 4:33 AM Terry Cox <terry@...> wrote:The intention is to get the Data Scientists out of the habit of working in isolation on their laptops for months and focus them on being regular and iterative collaborators to an MLOps team in a distributed production-capable environment.In practice, most of the DS team seem keen to maintain the service component of their ML solution and as a result are learning very rapidly about the practical challenges of scaling distributed solutions effectively.Terry CoxBootstrap Ltd+44 (7797) 828891+44 (845) 287-2345On 7 Jan 2020, at 01:12, Michael Neale <mneale@...> wrote:That is pretty cool Terry.With the training repo - is the idea that datascience type persons work in that, play with their data/model/notebooks, and when ready commit or issue a pull request to the *training* repo? do they then look at the PR in the service repo to test the real thing out? what is the workflow between data science experts and the engineers that work on the service/app that you see out there?On Fri, Dec 20, 2019 at 5:41 AM Terry Cox <terry@...> wrote:As discussed in the SIG meeting, here is a link to a quick demo video of MLOps within Jenkins-X. Things are developing fairly quickly and we plan to have a broad range of template projects in our public repo in the new year.
https://youtu.be/AqL_ME7BM6UTerry CoxBootstrap Ltd
--Regards,Michael Nealetwitter: @michaelneale, skype: michael_d_nealeCell: +61 423175597 (Australia)Cofounder @ CloudBees--Regards,Michael Nealetwitter: @michaelneale, skype: michael_d_nealeCell: +61 423175597 (Australia)Cofounder @ CloudBees