The Last Domino, Low-Code Data Science Falls In Line

The Last Domino, Low-Code Data Science Falls In Line

Low-code and no-code (LC/NC) software is pervasive. But although these accelerating approaches to software application development and data science have been touted as the answer to Artificial Intelligence (AI) democratization, no-code drag-and-drop breakthrough innovations have been few and far between. Some argue that low-code is headed in the same direction because it divides the ‘talent mosaic’ (or total pattern of the people universe with different skillsets) of data scientists and other business analysts.

Aiming to break some of these still-establishing conventions is Domino Data Lab with its Domino Code Assist (DCA) technology.

Described as a code-first approach (i.e. low-code, but really created for developers working with base-level coding tools that resonate with the way command-line interface technologies function) to democratizing access to data and analytics, the engineers behind this software promise it provides a path for organizations to transform their data science talent strategies.

To involve more business analytics professionals in data science and Machine Learning (ML) projects, DCA claims to be able to ensure that anyone (and the company means business users too) can generate code (Python, R) for common data science and analytics tasks based on simple selections in a GUI.

Part of Domino Data Lab’s flagship MLOps platform, Nick Elprin, co-founder and CEO at Domino Data Lab suggests that DCA may mark something of a watershed moment for low-code because it is rooted in the same coding languages used by most Artificial Intelligence (AI) professionals at companies.

“By delivering a common platform with access to low-code and professional-grade tools, Domino says it is removing additional barriers to writing code and further breaking down the myth that coding isn’t accessible by enough people to be a requirement in a large data science organization,” notes Elprin and team.

Sexiest job around

Described by the Harvard Business Review as ‘the sexiest job of the 21st century’, data science roles are projected to have a growth rate of nearly 28% by 2026, according to the U.S. Bureau of Labor Statistics. DCA enables Chief Data Officers (CDOs) and Chief Data Analytics Officers (CDAOs) to involve more analytics professionals in data science and machine learning projects using a code-first strategy.

By generating the code which helps business analysts learn and grow, DCA offers them an opportunity to learn coding skills and helps an organization develop new data science talent from within, all while improving efficiency of both business analysts and data scientists.

Domino’s code-first approach to low-code data science allows business analysts to generate code rather than using a black-box tool or proprietary drag-and-drop interface. This allows data prep, dashboards and models to be transparent and portable while enabling business analysts to collaborate with advanced data scientists in the same languages and environments scientists work in.


“Python is the new Excel and Domino Code Assist is about pushing the adoption of code-first data science in a way that has not been conceived of before,” said Sean Otto PhD, director of analytics at AES Corporation. “By enabling both our advanced and low-code team members with a common platform to deliver AI and ML with Domino, we can accelerate the innovative use of our data across a wide variety of technical and non-technical data practitioners and data-adjacent parts of our business.”

A code-first pyramid

With DCA, enterprises can pursue a code-first ‘pyramid’ or (as referenced above) a ‘mosaic’ data science talent strategy. What this means in less flowery language is that organizations can leverage a select group of skilled experts and larger numbers of less advanced members who still write some code. With everyone working in code on the same platform, organizations maximize the value of experts, while enabling and empowering analytics professionals at all levels.

“No-code solutions with drag-and-drop interfaces are appealing at first because they promise an easy path to data science, however, their limitations can prove frustrating for unlocking the full value of data and the inner workings of analytics solutions are often a ‘black box.’ This leads leaders to skepticism and mistrust around insights for business-critical decisions,” said David Stodder, senior director of research at research and education for data analytics company TDWI. “Building a code-first talent pyramid connected by shared data science programming languages improves data science collaboration, the mentorship of newer data scientists and ultimately increases workload capacity while accelerating transformation into a true data-driven organization.”

By empowering a broad range of data science and analytics professionals to unlock the power of data in a single platform, DCA may be useful in terms of helping CDOs, CDAOs and their staff address three key elements of democratizing AI.

AI’s 3-democratization factors

First, DCA helps address the ‘cold start’ problem by reducing time-to-insight for everyone. Business analysts can use wizards in DCA to automatically generate standard Python and R code for data ingestion, preparation, visualization and app creation tasks. They can also review other models and analyses as the starting point for their work and to help learn the necessary commands and syntax. For data scientists, DCA boosts productivity by eliminating the need to write code and remember the precise syntax for repetitive tasks.

Second, DCA exposes low-code business analysts to well-written code. It generates standard, editable Python and R code for common tasks. Business analysts get a repository of existing work for use in helping ensure that their work is consistent with the rest of the team. They can also easily collaborate with more experienced peers and mentors. All team members can save snippets of code as wizards to share across the organization, increasing code re-use, best practice adoption and standardization efforts.

Third, DCA helps align all data science work with broader governance, prioritization and review processes. Data-savvy analysts and other non-data scientists can safely gain the professional data science skills they need. With DCA, less-experienced analysts’ can have someone more seasoned review their work to ensure that the datasets are being used appropriately, the model was trained correctly, the project is scoped properly and aligned with organizational priorities.

Will AI and ML now be democratized overnight?

Obviously not, but there are some key pointers here such as lower-code tools that work like pure DNA coding tools and business functions in the AI space that work like businesspeople would expect them to. We can still be smarter, so let’s democratize, normalize and (when it comes to your hands in the post-pandemic aftermath) still also sanitize too.


Leave a Reply

Your email address will not be published. Required fields are marked *