Nick Elprin, Co-Founder & CEO
With digitization becoming the new norm in businesses worldwide, it is data that has taken the reins to drive organizational success. Tapping into the power of data science to stay ahead in the game is now more crucial than ever. As such, organizations are continually looking to streamline their data science workflows for mining the trove of data at their disposal and glean valuable insights that can power their business. While data science-generated models are at the core of accelerated business growth, many companies are missing the mark in creating a true model-driven business. For pharmaceutical and life sciences organizations, the challenge is more critical due to complex models, high expenses involved, and lack of collaboration, among many other factors. Fret not! San-Francisco based Domino Data Lab is setting new standards in the pharmaceutical analytics space by enabling data science to become a first-class organizational practice that companies of any size can run effectively at scale. “At Domino, our mission is to help companies put data science at the heart of their business,” begins Nick Elprin, co-founder and CEO of Domino.
Unleashing the Power of Data Science in Pharma and Life Sciences
Domino stands on a belief that, amid all the hype around data science, machine learning, or AI, the core unit of value is the model. As demonstrated by companies like Amazon and Netflix, model-driven businesses already or will come to dominate their markets. When it comes to model management in life sciences organizations, Elprin stresses that establishing an efficient and productive data science practice is no small feat. The research and development (R&D) process for life sciences companies is not just expensive and time-consuming but also requires extensive collaboration across data science teams. In addition, the nature of all life sciences work is experimental and collaborative; models must be constantly tracked, retrained, and iterated on, to reflect the changing data and other factors that lead to model drift. Elprin touches upon how, unlike software engineering or data management that depends on defined requirements, models require a research-based approach comprised of constant exploration, iteration, and agility. “Models can change their behavior as data in the real world changes. It’s important for data science teams to know about these behavior changes when they happen,” he adds. Actively working together with large and distributed data science teams entails major complications owing to the diverse skill sets and preferences for specific tools and technologies. Besides, it is an uphill task to promote visibility into the progression of projects, evolving experiments within the projects, and even the ROI achieved by data science teams.
To help companies navigate the labyrinth of complexities, Domino has introduced a novel open data science platform that is focused on helping data science teams accelerate the development and delivery of high-impact models while increasing collaboration, reproducibility, and reusability across the organization. The uniqueness of Domino’s platform stems from its openness and flexibility in terms of usage.
At Domino, our mission is to help companies put data science at the heart of their business
“Data scientists within the same organization often need different tools and languages for their research. Domino is an open platform, so it supports the tools, languages, and data sources that different teams need, now and in the future,” explains Elprin. For instance, if a particular data science team builds models on Python in Jupyter notebooks while the others use RStudio, they can collaborate on all of that work through Domino’s platform. Elprin mentions, “We can deploy the platform on-premise or in the cloud, and we connect it with a client’s existing tools and compute environment.” The flexibility, scalability, and elasticity of the cloud deployment prove immensely beneficial, specifically for R&D teams, wherein Domino’s platform can be integrated with any public cloud, such as AWS. Domino enables data scientists to spin up large or specialized hardware with one click, streamlining the resource provisioning process. The platform automatically spins resources down when they are no longer in use, preventing runaway costs.
Most importantly, serving as a single system of record, the Domino platform helps data science teams build institutional knowledge. The existing research, work, and datasets used for each of the runs and experiments can be tracked, versioned, and made available for review on the platform, enabling collaboration among the teams, particularly for highly-regulated industries with compliance and inspection support. The platform automatically versions not just the code but also the datasets used for each run, which ensures that the entirety of the experiment including the data, code, environments, discussion threads, and all the necessary artifacts are available for reproducibility, ensuring work is never lost, can be shared and built upon. In essence, the platform is a single source for collaboration, which saves team members the time invested in reinventing the wheel by cleansing and prepping the data before use. Consequently, with the Domino platform, users can beef up productivity which increases their business value to the organization—they can build and deploy more models that perform better in a shorter period of time, ultimately helping their business become model-driven.
Data Science as a Key Source of Competitive Advantage
While industry peers aim at simplifying data science for business analysts and software developers, Domino fosters the use of sophisticated tools crafted for assisting data scientists in managing their work. Elprin emphasizes, “The most valuable, complex work of data scientists cannot be streamlined using simple drag and drop tools. With its novel data science platform, Domino facilitates a culture of collaboration for sharing insights, enabling teams of skilled data scientists to iterate quickly at scale across enterprises.”
Domino takes pride in serving the world’s model-driven companies with its open technology data science platform. The company’s unified data science platform proved to be of great value to Bristol-Myers Squibb (BMS), a U.S.-based pharmaceutical company. Being at the forefront of cancer research, BMS deploys a strategic large-scale approach to data science and is committed to improving treatment responses and developing the cure. BMS is ambitious about their cancer research program, discovering potentially predictive biomarkers in tumors. The scale, complexity, and velocity of their data and analyses made tracking and sharing their work through traditional means labor intensive and inefficient. With the help of Domino’s platform, BMS’s translational bioinformatics team can now successfully test tens of thousands of hypotheses with complete reproducibility to accelerate their fight against cancer. Not just that; the research team is able to seamlessly track and trace the reams of genomic data aggregated from thousands of tumor samples. The platform acts a single source of record for researchers to find the exact snapshot of the code, data, or environment they are looking for, along with the comments or annotations made by the team, in turn, revving up the testing process.
Heading Toward a Promising Future
Following the latest release of the Domino platform—version 3.3— the company is realizing the vision of supporting the entire model management life cycle comprising model governance, production, development, and the underlying technology. “As the data science market matures and the hype around artificial intelligence, data analytics, and data science intensifies, we remain hyper-focused on working with organizational teams and practitioners to help them become more productive in developing and iterating models,” informs Elprin. The company continues to empower model-driven organizations to institute data science across enterprise-wide disciplines with their latest releases. With three new breakthrough capabilities—Datasets, Experiment Manager, Activity Feed—Domino is well on its way to help data science teams expedite the development and delivery of models through increased collaboration, reproducibility, reusability.
Domino will also be hosting the Rev summit next month, in New York City, focused on providing practical guidance to data science leaders for building successful data science teams, technical use cases, and best practices. Participating in the summit will be notable industry luminaries such as Nobel Prize winner Daniel Kahneman along with data science and engineering leaders from Slack, Netflix, WeWork, Google, and more. Elprin mentions that while other data science, AI, or machine learning conferences are directed at individual practitioners, through the Rev summit, Domino aims to shed light on the evolving role of data science leaders and how they can spearhead high-performance teams to solve bigger and more complicated problems.
Extending its geographical footprint, Domino will continue its expansion into Europe and India along with a team in London as well as setting up a second U.S. office in New York City. The company is geared to acquire new customers in the Asia-Pacific region, with an aim to demonstrate how model-driven businesses with robust data science infrastructure and platforms do not face an existential threat and can drive breakthroughs and operational improvements