Project to Production Challenge: AskAnna wants to solve this conundrum for ML Engineers and Data Scientists

In the AI Startup of the Week, the editorial staff of ai.nl is featuring promising AI startups, their innovations, solutions and challenges. In this fourteenth episode, we are taking a look at AskAnna, a Dutch AI startup that makes it easy for data scientists, ML engineers, and other professionals to track, trace, and share their data science work.

One of the biggest hurdles faced by data scientists is the amount of time it takes for them to work from development into production. In the past few years, we have seen how the demand for high performance computing has gone up in the field of machine learning, especially in the form of accelerators that speed up training of machine learning models and deployment.

However, a Dutch startup named AskAnna wants to solve this data science problem with a data science product of its own. As a software platform, AskAnna wants to help people kickstart their data science project without worrying about the speed. It does this by offering a set of tools that allows users to run their jobs in the cloud, including MLOps solutions.

AskAnna: helping AI professionals with their work

AskAnna is solely developed with the intent of solving the problem faced by data scientists, ML engineers, and other AI professionals who begin by creating their own model and try to then bring that to work. It is a known fact within the AI industry that a number of people drop their work because it is never likely to go into production.

AskAnna wants to solve that attrition with its interesting approach to transition of a code from development to production. It begins by identifying the workflow which starts by creating a model, bringing it to work, and then briefing the software engineers. This can be a slow process and AskAnna serves as a tool to improve your efficiency, collaboration, and speed.

AskAnna is essentially a web platform where users can bring their own model, collaborate, and share their data science project. It also offers data science version control, which acts as a centralised dashboard for users to check how a result was generated. Within seconds, users can see the input being used, the results being generated, and the environment or code the job was running.

The USP of AskAnna is its ability to offer all this data within seconds and on just one page. In fact, speed is of such essence that AskAnna can be set up and running in just 15 minutes. With AskAnna, developers can create their data science DevOps infrastructure they want using Yaml.

This speed is a far cry from the current limitations where setting up the infrastructure to operationalise could take days with DevOps. This ease of access and ability to set up infrastructure also results in data scientists getting their control over MLOps back.

Ease, collaboration and automation

The three core pillars of AskAnna have to be its relative ease of use, collaboration, and automation. The ease of use begins with the easy and flexible templates offered by the service. This allows data scientists, ML engineers, and AI professionals to stop doing the same thing over and over again.

There is an option to reuse the code, models, and even your favourite project structure. This feature is extremely attractive for people working in teams and looking to structure their project in a way that is understandable by each and every team member. AskAnna also allows users to quickly label their runs so they get a logical way to look at their successes.

AskAnna also offers the infrastructure necessary to compare your runs and spot the differences between them. Once you begin tracking your runs on AskAnna, it becomes easier to trace them and even compare them. All these automation features are supported by a sharing option, which is similar to sharing an image.

You take your work on AskAnna and share it with a team member or a collaborator by simply sharing a link. “You can share every run with a unique link, so your teammate can directly open the result you are referring to. And because you can invite unlimited people, you can share it with everyone you want,” the company promises on its website.

AskAnna: how much does it cost

AskAnna is still running in beta and those interested can sign up by clicking on this link. All the features are free till the service is in beta but once AskAnna comes out of beta, the company plans to offer three different types of tier. AskAnna also confirms that tracing and comparing your run is not yet available in beta state but will be added soon.

  • Free tier: 1 workspace, 2 projects, 3 users, 100 minutes of runtime with 1 CPU and 2 GB memory, 3 months history, 10 GB storage, and chat support
  • Basic tier at €250 a month or €2,500 annually: 1 workspace, unlimited projects, unlimited users, ability to run all day with 1 CPU and 2 GB of memory, unlimited history, 100 GB storage, chat and video support
  • Tailored tier starting from €2,500 a month: more than 1 workspace, unlimited projects, unlimited users, on premise with ability to run AskAnna in your own cloud, access to all your history including the runs you removed, 1 TB storage, hands-on support

AskAnna: what about workspaces and projects

If you are a machine learning engineer, data scientist, or an AI professional figuring out your work, then you must already be thinking about workspaces and projects available on AskAnna. Even though it is still in beta, AskAnna does have some projects ready for users to explore.

There is Python 2021, which is essentially the solutions of the Advent of Code 2021 edition written in Python. AskAnna also offers access to the Moby Bikes project, which aims to demonstrate several components relevant to a data science project and is available via two data pipelines designed to access SmartDublin Moby bikes data and weather data.

There is a TensorFlow project that acts as a demo on how to train a neural network to classify images of clothing. This project depicts how AskAnna can be used to run the TensorFlow environment, train a neural network, save files as artefacts, track variables and metrics, and save the trained model as the run’s result.

The projects and workspaces available on AskAnna may seem limited at this moment but it is an ambitious project aimed to reduce the time it takes for AI professionals to bring their project to production. If it succeeds in that endeavour, then it could emerge as the mightiest tool in an AI professional’s programming toolkit.

1440 1056 Editorial Staff

Editorial Staff

My name is HAL 9000, how can I assist you?
This website uses cookies to ensure the best possible experience. By clicking accept, you agree to our use of cookies and similar technologies.
Privacy Policy