From f0881c7f2c54fa06dd229792153fac458208ce6c Mon Sep 17 00:00:00 2001 From: Alessandro Lavelli Date: Fri, 4 Oct 2024 17:13:37 +0200 Subject: [PATCH 1/3] WIP on dev-project --- md-docs/user_guide/project.md | 17 +++++++++++++++++ mkdocs.yml | 1 + 2 files changed, 18 insertions(+) create mode 100644 md-docs/user_guide/project.md diff --git a/md-docs/user_guide/project.md b/md-docs/user_guide/project.md new file mode 100644 index 0000000..51fd152 --- /dev/null +++ b/md-docs/user_guide/project.md @@ -0,0 +1,17 @@ +# Project + +A Project is the secondary organizational entity in the ML cube Platform hierarchy. +A Project groups together a set of artificial intelligence algorithms that share a common goal expressed by a set of KPIs. +For this reason, it is composed of several [Tasks]. + +Users in the [Company] can have access to one or more Projects according to their roles. + +When a Project is created, the [User] specifies its *name*, *description*, and selects the *default storage policy*. +Storage policy defines the default behavior the Platform follows to store data that are shared with it. +Indeed, data can be shared via direct upload or remote cloud data source. +In the first case, data are copied in the ML cube private cloud storage, while, for the second one there is the possibility to do not duplicate data keeping them only in the source cloud + + +[Company]: company.md +[Tasks]: task.md +[User]: user.md \ No newline at end of file diff --git a/mkdocs.yml b/mkdocs.yml index 1e59333..ad1fa9d 100644 --- a/mkdocs.yml +++ b/mkdocs.yml @@ -107,6 +107,7 @@ nav: - User Guide: - user_guide/index.md - user_guide/company.md + - user_guide/project.md - Modules: - user_guide/modules/index.md From eac7faba0d688ca0a8fdcd89bcc1f0daa0646dba Mon Sep 17 00:00:00 2001 From: Alessandro Lavelli Date: Mon, 7 Oct 2024 09:05:51 +0200 Subject: [PATCH 2/3] Project page --- md-docs/user_guide/project.md | 54 +++++++++++++++++++++++++++++------ 1 file changed, 45 insertions(+), 9 deletions(-) diff --git a/md-docs/user_guide/project.md b/md-docs/user_guide/project.md index 51fd152..4dfd0de 100644 --- a/md-docs/user_guide/project.md +++ b/md-docs/user_guide/project.md @@ -1,17 +1,53 @@ # Project -A Project is the secondary organizational entity in the ML cube Platform hierarchy. -A Project groups together a set of artificial intelligence algorithms that share a common goal expressed by a set of KPIs. -For this reason, it is composed of several [Tasks]. +A Project is the secondary organizational entity in ML cube Platform. +A Project groups together a set of artificial intelligence algorithms that share a common goal measured by a set of KPIs. +For this reason, it is composed of several [Tasks] that collaborate to reach the Project's goal. -Users in the [Company] can have access to one or more Projects according to their roles. +Users in the [Company] can access to one or more Projects according to their [roles]. -When a Project is created, the [User] specifies its *name*, *description*, and selects the *default storage policy*. -Storage policy defines the default behavior the Platform follows to store data that are shared with it. -Indeed, data can be shared via direct upload or remote cloud data source. -In the first case, data are copied in the ML cube private cloud storage, while, for the second one there is the possibility to do not duplicate data keeping them only in the source cloud +## Creation +When a Project is created, the [User] specifies its *name*, and *description*, and selects the *default storage policy*. + +*Storage Policy* defines the default behavior the Platform follows to access data that are shared with it. +Indeed, data shared with ML cube Platform can either be duplicated and stored in ML cube private cloud storage or stay only on customer's cloud and accessed as a remote data source. + +## Demo Projects + +To better explore ML cube Platform modules and features, it is possible to create *Demo Projects* that are not taken into account by subscription quotas. +ML cube Platform provides different Demo Projects that cover all the possible use cases (regression, classification, text data, image data, RAG, object detection and so on). +To create a Demo Project, you need to check the "Demo Project" checkbox and select the one you prefer. + +## KPI Monitoring + +A Key Performance Indicator is a measure of performance over time for a specific objective, while artificial intelligence algorithms try to minimize their loss function, artificial intelligence based solutions and applications look at KPIs. +Therefore, it is essential to monitor Project's KPIs along with algorithm performance to have a complete view of the current situation. + +ML cube Platform offers the possibility to upload Project's KPIs to monitor them via drift detection algorithms. +That enables the detection of potentially dangerous trends in what really matters from the business point of view. +*KPI Monitoring* page in the Project sidebar shows the registered KPIs, their trends and drift events ML cube Platform detected during time. + +## Integrations + +ML cube Platform is part of the artificial intelligence and cloud ecosystem and provides connectors to interact with Cloud Providers and MLOps solutions. +The *Integrations* page allows to create and manage credentials that will be used by the Project's [Tasks]. + +For instance, if data are stored in a Google Cloud Storage bucket, adding the Google Cloud Platform credentials with the right permissions, allows ML cube Platform to read data from it. + +Another example is to trigger a Sage Maker pipeline to retrain your artificial intelligence model with a dataset provided by ML cube Platform. +In this case, you can create Amazon Web Services credentials with permission to create an event on Amazon Event Bridge. +See the [Integrations] page for more information about credentials setup, data sources and retraining triggers. + +## Jobs Monitoring + +Operations like sharing data to ML cube Platform, submitting the creation of a retraining dataset or reports like RAG evaluation, trigger the execution of asynchronous pipelines in ML cube Platform cloud infrastructure. +Each pipeline is associated with an identifier named *job id* that can be used to monitor its execution status. +This monitoring can be done both from Web App in the *Job Status* page and, with provided SDKs allowing automation. +A job failure can be either due to bad requests or internal errors, you can check the error message information via the same page. [Company]: company.md [Tasks]: task.md -[User]: user.md \ No newline at end of file +[User]: user.md +[roles]: rbac.md +[Integrations]: integrations/index.md \ No newline at end of file From 7ecb6e0aa05733d26236db5a646d9f478b65aa91 Mon Sep 17 00:00:00 2001 From: Alessandro Lavelli Date: Mon, 7 Oct 2024 11:43:02 +0200 Subject: [PATCH 3/3] fixes --- md-docs/user_guide/project.md | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/md-docs/user_guide/project.md b/md-docs/user_guide/project.md index 4dfd0de..b849d31 100644 --- a/md-docs/user_guide/project.md +++ b/md-docs/user_guide/project.md @@ -8,7 +8,7 @@ Users in the [Company] can access to one or more Projects according to their [ro ## Creation -When a Project is created, the [User] specifies its *name*, and *description*, and selects the *default storage policy*. +When a Project is created, the [User] specifies its *name*, *description*, and selects the *default storage policy*. *Storage Policy* defines the default behavior the Platform follows to access data that are shared with it. Indeed, data shared with ML cube Platform can either be duplicated and stored in ML cube private cloud storage or stay only on customer's cloud and accessed as a remote data source. @@ -21,7 +21,8 @@ To create a Demo Project, you need to check the "Demo Project" checkbox and sele ## KPI Monitoring -A Key Performance Indicator is a measure of performance over time for a specific objective, while artificial intelligence algorithms try to minimize their loss function, artificial intelligence based solutions and applications look at KPIs. +A Key Performance Indicator is a measure of performance over time for a specific objective. +While artificial intelligence algorithms try to minimize their loss function, artificial intelligence based solutions and applications look at KPIs. Therefore, it is essential to monitor Project's KPIs along with algorithm performance to have a complete view of the current situation. ML cube Platform offers the possibility to upload Project's KPIs to monitor them via drift detection algorithms. @@ -43,7 +44,7 @@ See the [Integrations] page for more information about credentials setup, data s Operations like sharing data to ML cube Platform, submitting the creation of a retraining dataset or reports like RAG evaluation, trigger the execution of asynchronous pipelines in ML cube Platform cloud infrastructure. Each pipeline is associated with an identifier named *job id* that can be used to monitor its execution status. -This monitoring can be done both from Web App in the *Job Status* page and, with provided SDKs allowing automation. +This monitoring can be done both from Web App in the *Job Status* page and, with specific SDKs method allowing automation. A job failure can be either due to bad requests or internal errors, you can check the error message information via the same page. [Company]: company.md