PureML x NayaOne Partnership Announcement
PureML x NayaOne Partnership Announcement
PureMLLogo

MLOps Wiki

This MLOps wiki provides a set of easy-to-understand explainers to assist anyone understand what the different terms of MLOps are, why they're important, and how they're managed in the ML lifecycle.

Data Monitoring

Data Integrity

A data pipeline encapsulates a comprehensive framework of processes and tools that orchestrate, streamline, and transform data from its source to a designated destination. Operating as a conduit of data handling, data pipelines automate complex operations, optimizing the flow of information across systems.

Data Pipeline

A data pipeline encapsulates a comprehensive framework of processes and tools that orchestrate, streamline, and transform data from its source to a designated destination. Operating as a conduit of data handling, data pipelines automate complex operations, optimizing the flow of information across systems.

Data Segments

Data segments facilitate efficient data analysis by partitioning and clustering similar data based on the chosen parameter or filter, enhancing the analytical process.

Training-Serving Skew

Training-serving skew refers to the discrepancy between a machine learning model's performance during training and its performance when deployed for real-world predictions, often arising from data handling differences and distribution shifts.

Drift

Concept Drift

Concept drift arises when real-world changes lead to shifts in variables, rendering previously learned patterns obsolete. This phenomenon detrimentally impacts the predictive capability of machine learning models, as their acquired knowledge becomes invalid.

Data Drift

Data drift, a phenomenon of paramount significance, emerges when the distribution of features within training and serving data shifts, even as the fundamental relationship between input and target variables remains unchanged.

Model Drift

Model drift, a nuanced facet in the realm of machine learning, pertains to the erosion of an ML model's predictive efficacy due to transformations in the digital landscape and consequential shifts in variables, including data and concepts.

Model Monitoring

ML Monitors

Machine Learning (ML) monitors serve as indispensable tools that play a vital role in scrutinizing model performance, pinpointing issues in prediction quality, and swiftly alerting stakeholders to real-time developments.

Model Behavior

Model behavior encompasses both global and local interpretations, encompassing features, target predictions, and insights derived from data segments.

Model Fairness

Model Fairness in the realm of machine learning (ML) encapsulates the quality and impartiality of models, directly linked to equitable allocation and the quality of services rendered. This aspect of ML is pivotal in ensuring that AI technologies are just, unbiased, and inclusive.

Predictions

Predictions, the outcome of a machine learning (ML) algorithm fed with historical input data, play a pivotal role in generating actionable insights and informed decision-making.

ML Ops

CI/CD for ML

Enabling the seamless orchestration of source code development, automated testing, and continuous delivery, CI/CD for ML pipelines stands as a foundational practice within modern software engineering.

Data & Model Versioning

In the intricate landscape of machine learning (ML) development, the concept of data and model versioning emerges as a cornerstone. This practice, akin to the meticulous archiving of progress, involves uniquely naming multiple iterations of ML models deployed at distinct stages.

Feature Management

A feature store serves as a centralized repository within an organization for curated features.

LLMOps

LLMOps represents a progressive facet within the MLOps domain, singularly dedicated to the operationalization of expansive language models (LLMs).

ML Diagnostics

In the realm of machine learning, the concept of diagnostics takes on a pivotal role in identifying challenges and refining models at various junctures of development.

ML Governance

ML Governance serves as a foundational internal framework within enterprises, orchestrating the control, execution, and oversight of ML development endeavors while strategically assigning roles and responsibilities.

ML Model Metrics

ML model metrics serve as the yardstick for assessing the effectiveness of machine learning models. These metrics play a pivotal role in the machine learning pipeline, offering validation and insights into model performance.

ML Reproducibility

ML reproducibility stands as a pivotal facet in the landscape of machine learning, ensuring that the processes, insights, and outcomes generated in the past can be accurately replicated in the present.

ML Scalability

Machine learning (ML) scalability entails empowering ML models to seamlessly handle vast datasets and execute numerous computations in a manner that is both economical and time-efficient.

ML Stack

The ML stack serves as a comprehensive reference model, outlining the essential infrastructure components crucial for constructing, training, deploying, and scaling machine learning systems.

MLOps Tools

MLOps tools are the linchpin of modern machine learning endeavors, simplifying intricate development processes and ushering in a new era of maintainability and auditability for ML experiments.

ModelOps

ModelOps represents a comprehensive strategy geared towards facilitating the seamless operationalization of machine learning (ML) models, thereby delivering anticipated business value to enterprises.

Pipelines & Processes Automation

Automated Machine Learning (AutoML) serves as a conduit for streamlining the entire machine learning pipeline, encompassing data extraction, processing, and storage. This orchestrated automation brings forth scalability, effectiveness, and increased productivity to the forefront.

Machine Learning

Managing AI Artifacts

AI Artifacts constitute the invaluable outcomes derived from the training process, bearing significance throughout the various phases of the machine learning (ML) lifecycle.

Ground Truth

In the realm of machine learning (ML), the concept of ground truth holds paramount importance. It refers to factual data that is either observed or computed and can be subjected to objective analysis within a specific ML use case.

ML Lifecycle

The machine learning lifecycle embodies a sequential series of steps integral to data science projects. These steps form a cyclic process, each carrying significant weight and executed in a meticulously defined order.

ML Models

Machine Learning (ML) models are intricate constructs designed to decipher patterns, infer insights, and make predictions from data.

Machine Learning

Machine learning stands as a dynamic subset of artificial intelligence, enabling computers to emulate human behavior without explicit programming and progressively refine their learning through real-world interactions.

MLOps

MLOps, a fusion of machine learning and operations, encompasses a suite of practices that orchestrate the deployment and maintenance of machine learning (ML) models in a production environment.

Model Monitoring

Model monitoring stands as a critical practice in the realm of machine learning (ML), focusing on the continuous assessment of ML models in production environments to preemptively identify potential issues within ML pipelines.

Natural Language Processing

Natural Language Processing (NLP), a domain within Artificial Intelligence, empowers computers to comprehend and interpret human-written languages, bridging the gap between complex computer languages and users.

Overfitting Vs. Underfitting

In the realm of machine learning, the concepts of model overfitting and underfitting hold paramount importance as they reflect the delicate equilibrium between capturing the essence of training data and generalizing well to new data.

Responsible AI

AI Audit

Introduces a meticulously orchestrated audit program aimed at educating senior executives, unveiling inherent risks, and establishing safeguard controls, all in the pursuit of sound AI governance.

AI Black Box

A notion prevalent in the world of machine learning, encapsulates the utilization of models that resist comprehension solely through scrutinizing their parameters.

AI Explainability

AI explainability stands as a cornerstone in bridging the gap between the complex outputs of machine learning algorithms and human comprehension.

AI Observability

Stands as a comprehensive and sophisticated approach aimed at unveiling the intricate workings of machine learning models throughout their lifecycle.

Bias in ML

In the realm of machine learning, the specter of bias looms as a critical challenge, demanding vigilance and ethical considerations. Let's delve into the intricacies of this phenomenon.

Ethical AI

Ethical AI establishes a comprehensive framework encompassing guidelines concerning individual rights, privacy, behavior manipulation, and non-discrimination.

Promoting Responsible AI

Responsible AI (RAI) represents a set of principles and methodologies aimed at the development, deployment, and scaling of artificial intelligence with a positive impact on both people and society.