Here’s a look at 10 data science and machine learning tools that solution and service providers should be aware of. 
The data science and machine learning technology space is undergoing rapid changes, fueled primarily by the wave of generative AI and—just in the last year—agentic AI systems and the large language models that power them.
Where data science and machine learning tools were traditionally targeted toward developing and supporting data analytics and predictive analytics systems, these tools today are increasingly supporting the development of AI agentic systems, according to The 2025 Gartner Magic Quadrant for Data Science and Machine Learning Platforms.
Another major trend, according to the report, is a shift away from point-specific tools towards “full-stack data and AI platforms” that encompass model development and lifecycle management, data science tasks, data pipelines and other chores. What’s more, these platforms are themselves incorporating LLMs and GenAI assistants “to enhance the data science workflow.”
As part of CRN’s 2025 Year So Far series, here’s a look at some of the hottest data science and machine learning tools in use today. Some of the following tools are relatively new to the market while others have been around for a while and recently updated. The list also includes both commercial products and open-source software. 
Dataiku’s flagship Universal AI Platform is one of the industry’s leading AI and machine learning platforms for data scientists. In April Dataiku debuted AI Agents with Dataiku, a new set of capabilities within the platform for creating and controlling AI agents at scale.
The platform supports the central creation of agents with Code Agent for data scientists and developers and the Visual Agent no-code option for non-technical business users. Capabilities include Managed Agent Tools for maintaining the quality and validation of tools used by agents and a GenAI Registry for strategic oversight of agentic use cases.
A key technology within AI Agents with Dataiku is the Dataiku LLM Mesh architecture to manage model access across all proprietary, open-source and cloud service large language models, according to the company. Dataiku Safe Guard defines and applies guardrails while Agent Connect centralizes agent access across an organization from a single interface.
For agent observability and performance monitoring, AI Agents with Dataiku provides Trace Explorer for visibility into agent decision making, Quality Guard to continuously evaluate and monitor agent performance, and Cost Guard for real-time usage tracking, budget enforcement and internal cost allocation. 
Anaconda is well known for its data science and AI platform for developers using the popular Python programming language.
The new Anaconda AI Platform, unveiled in May, is a unified open-source platform that Anaconda says provides a comprehensive system for streamlining machine learning workflows and building, training and deploying machine learning models.
The platform provides simplified development and governance controls to boost practitioner productivity and reduce organizational risks associated with open-source AI development, according to the company.
Features and capabilities include Anaconda AI Navigator for AI application development and experimentation with large language model, and the AI-powered Anaconda Assistant chatbot that assists with coding, debugging and data visualization. It also includes Conda Package Manager for managing packages and dependencies, curated “essential” ML libraries, and MLOps for automating model deployment and management. 
In May, agentic workforce platform developer DataRobot debuted syftr, an open-source framework that’s designed to help AI developers evaluate and identify performant agentic workflows for commercial use.
According to DataRobot, syftr “empowers AI practitioners to programmatically discover and implement the best combinations of components, parameters, tools and strategies for agentic use cases” and optimize them for accuracy, processing speed and cost.
Some of syftr’s capabilities include multi-objective search and Bayesian optimization early stopping mechanism.
The syftr software is currently available as a “permissively licensed” open-source project. An enterprise edition of syftr will be available this fall. 
Domino Data Lab’s flagship Domino Enterprise AI Platform is a machine learning operations (MLOps) system that helps organizations build and operate AI at scale. The platform provides a central hub for data science teams, according to the company, offering tools and infrastructure for managing the entire data science lifecycle, from initial exploration to model deployment and monitoring.
In June, Domino Data Lab launched a new release of the platform with new capabilities, including a unified system for productivity, governance and delivery, “turning fragmented initiatives into an AI factory” for trusted, repeatable outcomes.
The release also included a new Zero-to-AI service to catalyze proven AI cultural change within an organization. 
Hex provides a collaborative data science and analytics workspace where data teams and business users can share analytical results. The platform combines the capabilities of traditional data science notebooks with integrated AI assistance, data applications and reports, and advanced collaboration functionality, according to the company.
In January, the company introduced Hex Embedded Analytics, which allows developers to build the Hex technology into data products such as applications that need customer-facing analytics.
In May, Hex raised an impressive $70 million in Series C funding. 
MLflow is an open-source MLOps platform for managing workflows and artifacts across the machine learning lifecycle, according to the mlflow.org website, assisting machine learning practitioners and development teams in handling the complexities of the machine learning process.
MLflow 3.0, introduced on June 11, “isn’t just another feature update,” according to the 3.0 release announcement, but “fundamentally expands what’s possible” with ML tooling and addresses observability and quality challenges around GenAI deployment.
The new edition provides the LoggedModel1 entity to enable better organization and comparison of generative AI agents, deep learning checkpoints, and model variants across experiments. It also offers a new GenAI evaluation suite and enhanced model tracking for lineage support.
The MLflow project was originally created by data management platform giant Databricks, which contributed it to the Linux Foundation in 2020. Databricks offers a fully managed MLflow service on its own platform.
Today MLflow has more than 30 million monthly downloads and contributions from more than 850 developers worldwide, according to Databricks. 
PyTorch is a widely used, open-source machine learning library and framework for developing and training deep neural networks. It is known for its flexibility and ease-of-use for more intuitive model building and debugging, along with its dynamic computation graphs capabilities, according to the PyTorch.org website.
The most recent edition is PyTorch 2.7.1 released on June 4, according to GitHub. The new version includes support for Python 3.12 and optimizations for AOTInductor.
PyTorch 2.7.1 is part of the PyTorch 2 series that focuses on enhancing PyTorch’s performance and the user experience through compiler-level changes. 
At its Snowflake Summit 2025 in early June Snowflake unveiled Data Science Agent, an “agentic companion” that the company said boosts data scientists’ productivity by automating routine machine learning model development tasks.
Snowflake said Data Science Agent simplifies AI and ML workflows, democratizes users’ access to data across their businesses, and eliminates technical overhead – all through a natural language interface within Snowflake, according to the company.
Data Science Agent, soon to be in private preview, uses Anthropic’s Claude large language models to break down problems associated with ML workflows into distinct steps, such as data analysis, data preparation, feature engineering and training, according to Snowflake. The product creates fully functional pipelines using such advanced techniques as multi-step reasoning, contextual understanding and action execution. 
Tecton got its start developing a feature platform that streamlines the process of building, deploying and managing machine learning features. The company expanded beyond its machine learning roots in September 2024 with a new release of its platform that delivers contextual data to the large language models that power generative AI systems.
In February, the company debuted Tecton 1.1, the latest update to the platform, with added capabilities the company says makes it simpler for AI teams to build more sophisticated features, optimize infrastructure and improve model performance.
The release includes new API resources for accessing any third-party data source in real time, a new capability for more efficiently performing the calculations needed for real-time feature views to speed up transformations during online retrieval queries, and a number of performance enhancements in the core Tecton platform. 
TensorFlow is a popular open-source machine learning platform and software library for developing and deploying machine learning models—especially sophisticated deep learning models and neural networks—for AI.
TensorFlow 2.19 was released in March with a number of technical improvements including changes to the C++ API in LiteRT and bfloat16 support for tflite casting.
While PyTorch is generally seen as an alternative platform for small-scale machine learning development projects where model experimentation and quick editing are priorities, TensorFlow is generally viewed as best for large projects and production environments that require performance and scalability, according to the TensorFlow.org website.
TensorFlow is available under the Apache License 2.0.
