An Interview with Dr. Ramprasad, Matmerize CEO and Co-founder
Matmerize, Inc., a Polymer Informatics software company, has a modern approach to accelerate the development of polymers and formulations. It integrates polymer domain knowledge with advanced data science methods.
Matmerize has created a cloud-based materials design software, called PolymRize™.
This software is based on machine learning and can be securely accessed by polymer and formulations chemists via a web browser or API calls.
Also, it seeks to define the new industry standard for cost-effective accelerated polymer materials and formulation development by eliminating expensive and long trial-and-error cycles.
It utilizes Matmerize-proprietary or client-proprietary data to build predictive models to guide experimental design and optimization of materials.
Recently, we interviewed Matmerize's CEO & co-founder, Dr. Ramprasad. Dr. Ramprasad brings decades of pioneering academic computational materials engineering work, unparalleled experience and unique IP in this space. Let us hear directly from him about their new flagship product PolymRize™, and what they have in store for their users.
1. What is Matmerize and what makes you unique?
The development of new polymers and formulations to meet evolving application demands is nontrivial. Traditional materials developmental activities are largely reliant on trial-and-error efforts that consume enormous time and resources, as they do not fully exploit past and emerging data, intuition, and experience. The parameter space of polymer formulations is so vast that numerous new functional and optimal polymer chemistries and formulations are very likely awaiting discovery.
Matmerize seeks to modernize and define the new industry standard for
cost-effective accelerated polymer materials and formulation development. We have a pioneering unique presence at the intersection of polymer engineering and data science; and we have world-class credibility, unparalleled experience, and key IP in this space.
Matmerize provides data-driven software capabilities for rapidly predicting properties of materials formulations before they are made (and tested) at minimal cost, does virtual screening, and suggests materials that are most attractive for which experiments should be performed. Matmerize thus envisions a world that inherently adopts economies of scale and effort.
2. What is Matmerize’s core product/technology, and what is the benefit of using this product?
Matmerize is a recent spin-out from the Georgia Institute of Technology and has licensed state-of-the-art Polymer Informatics IP pioneered in the Ramprasad Research Group.
1 With this IP and world-recognized experience,
Matmerize has built PolymRize™,
an industry-grade cloud-based Polymer Informatics software platform to accelerate the design and development of polymers and formulations. PolymRize™ utilizes Matmerize-proprietary data or client-proprietary data to build predictive models using proprietary machine learning (ML) algorithms.
Once built, the predictive models may be used to:
- Rapidly predict the properties (with uncertainties) of new yet-to-be-synthesized polymers and formulations.
- Recommend candidates that meet target property requirements (using either automated design tools or a user-supplied candidate set) to inform the next round of physical experimentation.
Our clients have indicated that they could converge on a target formulation in one-third to one-fourth of their usual time and cost if they use PolymRize™, while also providing them the discipline to structure and manage their data.
(Click on Image to Enlarge)
3. How does Matmerize handle security and confidentiality with respect to client information, data, and models?
All user interactions with PolymRize™ are encrypted, and not visible to anyone outside of the client company
(including to the Matmerize team).
When a client utilizes their proprietary data to build predictive models using
PolymRize™:
- Their models are deployed and available for the exclusive use of only that client, i.e., other clients of Matmerize or Matmerize employees will not have access to those predictive models.
- The client-proprietary data used to build such models are also destroyed once the model has been built and deployed, so no one has access to those datasets as well.
- All queries, histories, searches, and screening results are likewise encrypted and not available to anyone by the user.
4. Do you see PolymRize™ as a competition to internal digitization efforts and chemists?
Matmerize (via PolymRize™) leverages and complements internal digitization efforts (that organizes historical and emerging proprietary data) by helping in building ML models using such data. PolymRize™ is thus
a digital partner to chemists and process engineers, rather than a competitor. It may also be viewed as a state-of-the-art benchmark to calibrate the quality of internal models and internal informatics initiatives.
PolymRize™ may be accessed by either using:
- A web browser or
- Application Programming Interfaces (APIs).
Some of our clients use API endpoints to directly integrate PolymRize™ with their internal digitization and simulation workflows. This feature allows for seamlessly forecasting the outcomes of a new planned experiment or simulation before these tasks are performed.
Overall, rather than compete with or duplicate the internal activities of chemists, process engineers, and computational materials scientists, PolymRize™ offers complementarity, efficiency, and support to accelerate rational materials optimization or discovery.
(Click on Image to Enlarge)
5. How do clients typically engage with Matmerize?
Our intent is to have our clients subscribe to the PolymRize™ platform (which comes at different usage tiers and with periodic tutorial and consulting support).
The subscription arrangement may develop in following appropriate steps:
- After an initial introductory conversation with Matmerize to learn about our capabilities and services, clients identify 3-5 user cases of interest that may be addressed using PolymRize™.
- Now, depending on the problems chosen and the informatics-readiness of the client, an initial engagement may occur either as a 3-4-week free trial of the platform, or as a 3- to 4-month project performed by Matmerize collaboratively with the client.
- After this, clients are in a suitable position to choose an annual subscription at the right usage tier level appropriate for their needs.
Companies are beginning to realize the value and opportunity presented by leveraging materials data. Matmerize seeks to take the logical next, but a transformative, step –
leveraging the data to extract knowledge and guide the next experiments toward cost-effective and intelligent materials design.
About Ramprasad
Dr. Ramprasad formed Matmerize in 2019 and is its CEO and co-founder. He has been a Professor of Materials Science and Engineering at Georgia Tech since 2018. Prior to joining Georgia Tech, he was a faculty member at the University of Connecticut, and prior to that, he was a Staff Scientist at Motorola, Inc. Dr. Ramprasad got his B.Tech from Indian Institute of Technology (IIT), Madras, India, an M.S. degree at the Washington State University, and a Ph.D. degree at the University of Illinois, Urbana-Champaign.
His area of expertise is in the development and application of computational and machine learning tools to accelerate materials discovery, as applicable to energy production, storage and utilization.
Prof. Ramprasad is a Fellow of the Materials Research Society, a Fellow of the American Physical Society, an elected Member of the Connecticut Academy of Science and Engineering, and the recipient of the Alexander von Humboldt Fellowship and the Max Planck Society Fellowship for Distinguished Scientists.