Large language models (LLMs) can generate fluent responses, but they do not automatically understand an organization’s operating environment, internal documents, technical standards, or security boundaries. This paper presents a practical foundation for building Retrieval-Augmented Generation (RAG) workflows in Python using LangChain. Guided by explanations and code excerpts, this paper follows the construction of a RAG pipeline from document loading and text preparation through chunking, vector embeddings, FAISS indexing, semantic retrieval, prompt construction, and connection to hosted or localized language models. Common chunking strategies are compared to show how document structure and context preservation affect retrieval quality. This paper also considers the use of small and medium-sized open-source language models for well-scoped and bounded tasks, larger models for broader synthesis, and the role of retrieval, source citations, evaluation, access controls, and human review in reducing unsupported responses. The resulting workflow provides a practical starting point for secure enterprise and research applications in 2026, 2027, and beyond.

The rapid expansion of cloud-hosted environmental datasets through open data initiatives has changed how climate and weather data are accessed and analyzed; however, challenges remain in efficiently interfacing with large-scale repositories for operational and analytical workflows. This study presents a cloud-native Python framework for scalable access, processing, and analysis of environmental datasets stored in object storage systems including Amazon S3, Microsoft Azure, and Google Cloud Storage. The framework integrates Xarray, Dask, and fsspec to enable lazy loading, parallel computation, and efficient subsetting of high-resolution, multi-dimensional gridded datasets. It implements metadata-driven access, chunked data retrieval, caching, and optimized indexing using cloud-hosted datasets including CMIP6 and the operational NOAA RTMA dataset. Case studies demonstrate on-demand time series generation, spatial subsetting, and interactive visualization workflows without requiring full dataset downloads. Performance considerations and trade-offs between data formats (NetCDF, Zarr, GRIB, GRIB2) are evaluated across use cases, data structures, and access patterns in both research and operational contexts. This work establishes a Python framework for integrating cloud-based environmental datasets into forecasting workflows, research applications, and decision-support systems, with extensions to AI-enabled workflows. It enables scalable machine learning data pipelines through efficient extraction, alignment, and processing of multi-resolution environmental datasets. This approach supports supervised and unsupervised workflows, including resolution enhancement and pattern identification, for climate and environmental AI applications.

Generative AI is rapidly reshaping how organizations search, reason over, and contextualize information, yet deploying these systems on private, confidential, and sensitive data introduces distinct architectural, performance, and governance challenges. This paper presents a practical roadmap for the 2026 enterprise generative AI stack, with emphasis on agentic AI architectures built on retrieval-augmented generation (RAG), vector embeddings, and secure model deployment. Core concepts including encoding, similarity search, and vector databases are introduced to explain how knowledge is stored, retrieved, and reused across AI agents, alongside API-driven patterns that enable tools for search, reasoning, and contextualization. The discussion contrasts proprietary large language models (LLMs) and open-source small language models (SLMs), highlighting trade-offs in output quality, performance, and deployment, while examining how ecosystems such as Hugging Face support localized inference and domain-specific adaptation. Implementation considerations are presented primarily in Python, with extensions to R and SAS® Viya®, focusing on secure, reproducible analytics workflows. This paper concludes with prompt engineering strategies and ethical considerations for responsible generative AI use on sensitive enterprise data in the life sciences, pharmaceutical, and healthcare domains.

Developers, data scientists, and analysts are increasingly leveraging open-source tools and libraries to integrate with and enhance their existing data engineering and analytical workflows. One of these integrations, built into SAS® Viya®, is its pre-configured Python runtime integration, PROC PYTHON, that gives SAS programmers access to Python's open-source data science libraries for processing, visualizing, and analyzing data alongside SAS procedures. This presentation demonstrates how to access and use Python libraries in Viya runtimes; understand data-handling in Python and SAS; build Python scripts with reusable methods that import, process, and analyze data; and execute those Python methods to export Pandas DataFrames as SAS datasets.

This paper presents a practical framework for building reliable, reproducible, and collaborative data science workflows using Git, GitHub, Data Version Control (DVC), and MLflow. It begins by introducing Git as the foundation for tracking code changes and GitHub as a collaboration layer for shared repositories, branching strategies, pull requests, and team-based development. DVC is then presented as an extension to version control that enables datasets, intermediate outputs, and analytical artifacts to be tracked, compared, and restored alongside code without storing large files directly in Git repositories. The paper demonstrates strategies and techniques for meaningful commit practices, managing work-in-progress (WIP) safely, reducing merge conflicts, and maintaining structured project histories across collaborative environments. Building on this foundation, MLflow is introduced as a lightweight experiment tracking and model management layer for machine learning workflows in Python, allowing teams to record training runs, fine-tuning parameters, evaluation metrics, model artifacts, and performance comparisons during iterative development. Together, Git, GitHub, DVC, and MLflow provide an integrated open-source ecosystem for managing code repositories, versioning data, and tracking model experiments across collaborative data science, machine learning, and analytics projects.

The democratization, growth, and wide-spread adoption of the Python programming language and its open-source libraries are empowering organizations, businesses, professionals, researchers, educators, and students alike to work directly with big climate data repositories in the cloud. This open-source movement, coupled with Open Data initiatives sponsored by large technology providers like Amazon, Microsoft, and Google is ushering in an age of data democratization from government agencies and organizations in the cloud. This paper provides a fundamental roadmap to help beginners of all levels, across all industries, and possessing any professional background to connect with, access, process, visualize, and conduct analysis on big climate repositories stored in the Amazon S3, Microsoft Azure, and Google Cloud Storage providers. This paper discusses the motivations and applications driving the adoption of cloud object storage services for climate datasets, including common file types, data formats, data archives, and examples of publicly available climate datasets stored in these cloud providers. Examples of these big datasets include remote-sensing and monitoring station repositories like CMIP6, GHCN, RTMA, CMORPH, and more. Several Python libraries and big data processing strategies are investigated to assist with accessing, visualizing, and efficiently handling these big climate datasets stored in cloud storage providers.

Organizations around the globe are truly facing a paradigm shift with the type of software, the quantity and availability of software technologies, including open source, and the creative ways the many technologies live, play, and thrive in the same sand box together. We’ll explore the many benefits, challenges, and opportunities with open source technologies in the 21st century. We’ll also describe the challenges facing user communities as they find ways to integrate open source software and technologies, handle compatibility and vulnerability issues, address security limitations, manage intellectual property and warranty issues, and address inconsistent development practices. Plan to join us for an informative presentation about the benefits, challenges, and opportunities confronting open source user communities around the world, including the application and current state of Python, R, SQL, database systems, cloud computing, software standards, and the collaborative nature of community in the 21st century.

Capable of accepting and mapping complex relationships hidden within structured and unstructured data, neural networks are built from layers of neurons and activation functions that interact, preserve, and exchange information between layers to develop highly flexible and robust predictive models. Neural networks are versatile in their applications to real-world problems; capable of regression, classification, and generating entirely new data from existing data sources, neural networks are accelerating recent breakthroughs in Deep Learning methodologies. Given the recent advancements in graphical processing unit (GPU) cards, cloud computing, and the availability of interpretable APIs like the Keras interface for TensorFlow, neural networks are rapidly moving from development to deployment in industries ranging from finance, healthcare, climatology, video streaming, business analytics, and marketing given their versatility in modeling complex problems using structured, semi-structured, and unstructured data. This paper explores fundamental concepts associated with neural networks including their inner workings, their differences from traditional machine learning algorithms, and their capabilities in supervised, unsupervised, and generative AI workflows. It also serves as an intuitive, example-oriented guide for developing Artificial Neural Network (ANN) and Convolutional Neural Network (CNN) architectures using Python's Keras and TensorFlow libraries for regression and image classification tasks.

Machine learning is experiencing a golden age of investment, democratization, and accessibility across all domains in the life sciences, natural sciences, and social sciences with applications to industry for business decision-making, risk management, consumer marketing, clinical trials, financial forecasting, security recognition, video remastering, digital twin simulations, and more. But what exactly is machine learning (ML)? How is it connected to Artificial Intelligence (AI)? And most importantly, how can data scientists, programmers, software engineers, and/or researchers start their endeavors into machine learning? This paper answers these questions, and more, by providing a roadmap to help navigate the complexities of machine learning in an application-oriented guide. This paper covers the main aspects of machine learning including supervised, unsupervised, and semisupervised approaches as well as deep learning. The roadmap for supervised machine learning starts with linear regression and progressively builds towards more complex and flexible algorithms with discussions about the advantages and disadvantages of using certain models over others. This paper discusses the real-world applications of both labeled and unlabeled data; supervised and unsupervised machine learning algorithms; overfitting and underfitting; cross-validation; and the importance of hyperparameter tuning to better fit algorithms to their data.

Google Earth Engine is a powerful cloud-based storage platform for accessing publicly available geospatial data from third party sources, including satellite imagery, geophysical, socioeconomic, climatological, census, and meteorological data measured over time for academic-use, personal-use, research, and business applications. Through a combination of beautiful visualizations and easy-to-implement Python code, users will be given the tools to conduct their own analysis with Google Earth Engine. Using the intuitive Python API, along with a suite of visualization packages and map-making libraries available for Python, this paper showcases methods for accessing, querying, extracting, and visualizing Earth Engine's spatiotemporal data to develop interactive maps. Optimized techniques permitting intensive spatiotemporal analysis on large, complex datasets are introduced through server-side operations in Google Earth Engine. By the end of this Paper, users will feel comfortable setting-up, configuring, and linking Earth Engine to Python, become acquainted with commonly-used formats for storing various types of spatial data, understand methods for querying, selecting, uploading, and exporting datasets from Earth Engine, effectively visualize high resolution spatiotemporal data using Python's Geemap package, and be able to conduct analysis using server-side operations to efficiently complete resource-intensive tasks.