Picominer® Edge

MLOps For Industrial Automation

Picominer aims to solve the software and process challenges in "AI/ML Model-To-Business Lifeclycle".

Picominer-v1.6-01.01-Picominer-Introduction
BIG PICTURE

The MLOps Problem

Majority (some reports say nearly 60%) of the AI/ML projects do not cross PoC (proof of concept) stage event though the data models are promising in the lab environment.

Unlike software development lifecycle, the data lifecycle is quite different. Usually, the cost of deploying AI/ML models are not factored in the initial estimates.  

Model-To-Business needs additional steps that adopt the models to target environment (application context) in the AI/ML project lifecycle. It is the motivation behind the design of Picominer.

Picominer-Edge-v1.7-04.00-vExpert-3
INDUSTRIAL AUTOMATION

Data applications

Managed MLOps

With Picominer Edge we build Model-To-Business data pipelines and offer managed services for to your target environment.

Integrations

Picominer Edge supports a variety of integrations and protocols with automation devices and shop floor MES/ERP applications.

Open source

Picominer Edge platform designs and reference implementation are released under open source licence. No vendor lock-in.

Team training

Our services include team training and support on MLOps, DevOps, Expert Systems etc. to suite your business environment.

SMART DECISIONS AT SHOP FLOOR

MLOps at the Edge

Picominer provides hosting of AI/ML models at the IIoT gateways. Picominer extends the capabilities of IIoT gateways with Ring, Star and Mesh topology.

Picominer-Edge-Presentation-v1.7
MOTIVATION

A lightweight AI/ML model serving platform for  IoT edge, automation and web.

SOLUTION
Binding the data models and expert knowledge for the target context.
BENEFITS

Customised MLOps. Incremental and continuous improvements. 

PROCESS

Picominer aims to solve the software and process challenges in "AI/ML Model-To-Business Lifeclycle".

PROJECT STUDY

Firstly, we will schedule meetings with customer
teams to set the scope.

DATA FLOW DESIGN

Identifying various data touch points to align with
the business objectives.

EXPERT SYSTEM

Data models will be subjected to the target
contextualisation process.

MODEL REGISTRY

Models are verified against target environment and a
model registry is prepared.

INTEGRATIONS

Integration (API) with MES/ERP and other
information applications.

Benefits that come with our engagement

Reference design

In-depth reference design, reference implementation and support for MLOps.

Flexible

Customise the MLOps to suite to the business. No opinionated framework.

DevOps

Versatile yet simple platform to run models at scale. DevOps may be used if needed.

Open source

All our reference implementations are released under open source. No vendor lock-in.

AI R&D

Access to our ongoing R&D on AI/ML reference designs and sources.

Team training

Get a structured training to the team on MLOps (online / offline as needed).

Innovation

Our approach is to enable the teams to innovate for the business needs.

Standards based

Incremental and continuous process improvement.