Ellf: Your virtual NLP engineer

Ellf gives your coding assistant like Claude Code superpowers for solving AI and Natural Language Processing (NLP) problems and a platform to execute data-centric projects, automated annotation, evaluation and more. It includes skills and knowledge for all parts of the NLP development lifecycle, including all our best practices from over 10 years of solving real-world problems in industry.

To keep your data 100% under your control, Ellf lets you plug in a powerful and fully data-private cluster that runs locally or in the cloud and makes it easy to execute annotation tasks, auto-annotation agents, training experiments and more, and collaborate on development with your team.

First steps

Once you’ve installed Ellf and connected it to your coding assistant, you can start by describing your project and asking it questions on strategy and next steps. If you already know what you want to do, you can also call the individual modules directly – for example, the ellf-project command starts the project planning module and ellf-annotate helps you design data annotation tasks.

To keep track of your project, Ellf will create detailed project plans. These are Markdown documents that are continuously updated and outline the end-to-end development plan, including step-by-step workflows and higher-level strategic goals.

NLP development typically involves running many different things: data processing, annotation, training experiments, evaluation and more, often iteratively and in parallel. To make this easier, Ellf can automatically start and monitor these processes and run them in a data-private cluster, either locally or on your own cloud account. This means that no data ever has to leave your servers.

The platform lets you view your tasks, actions, agents and services, as well as data assets on your cluster, and collaborate on them with your team. You can interact with Ellf and manage resources via the ellf command-line interface or visually in the app. The web UI also includes a command bar at the top that shows the CLI command corresponding to the current page or view, making it easy to switch between using the app and CLI. Your local coding assistant also uses the CLI to interact with Ellf.

For data annotation, Ellf ships with our popular and efficient annotation tool Prodigy, a range of built-in recipes for common tasks, as well as annotation agents to automate data development for you. Ellf can also help you select the right NLP components for your business problem and develop the label scheme before you scale up development.

Architecture

To keep your data private and allow flexibly scripting the application workflows in Python, Ellf implements a hybrid dual-cloud architecture, consisting of a web application and servers hosted by us, and a data processing cluster hosted by you locally or on your own cloud account and entirely under your control, similar to self-hosted runners in a CI system. The cluster stores and provides all code and data, which is executed and served directly from there.

Ellf generally stores records of data and other objects, as well as user and organization details, while your cluster stores the actual data and runs the code. Whenever data is shown or served in the web application or CLI, it comes straight from your cluster and doesn’t ever pass through servers hosted by us.

Human support

If you ever feel stuck or need human feedback, you can use the support module to send us your logs and have them reviewed by one of our NLP experts. Alternatively, you can also export your conversation log to a file and email it to us at support@ellf.ai.