Welcome to Pyristic Solutions

Merging Old & New

In many established manufacturing environments, we have to contend with old and entrenched paradigms and protocols. To take advantage of newer technologies to improve manufacturing services, we have to find a way to bridge the old and the new.

In this project, we designed and built a protocol on top of old customized FTP infrastructure to deliver applications containing sensitive data and code as docker images securely and automatically. Applications can be managed centrally and deployed anew or as updates to sites around the world, removing the manual burden of deployment and security from users. Furthermore, this platform integrates with existing company managed version control systems to reduce the need for manual input for application registration. Using modern elliptic curve cryptography we were able to protect sensitive application data in transit and ensure that the server being deployed to was cryptographically verified.

This project ensures that application updates are delivered quickly to remote sites while giving application developers the many benefits of containerized development and deployment.

Supply Chain Woes

The pandemic provided insight into the fragility of global supply chains. Years of centralization led to single points of failure that cascaded to failures across multiple industries. Many companies did not have proper ways to assess the risk of working with any particular vendor. This poses significant business risk should any supply chain disruptions occur.

This project focuses on mining existing large pools of data to provide metrics for determining the health of a vendor. Multiple analyses are performed to provide an aggregate score to manufacturing managers to understand how to plan their processes to reduce the risk of vendor misses. If a part doesn’t arrive on time, manufacturing capacity can develop a bottleneck, and the failure can cascade. We provide users with insight to plan according to the probability of when parts might arrive, thus improving manufacturing output and reducing the chance of bottlenecks.

Fingerprinting

One of the first steps after installing a new tool in most manufacturing environments is fingerprinting. It is important to understand the state of a tool immediately after installation. This is compared to post manufacturing data from the tool’s construction and subsequent performance when the tool itself begins performing its manufacturing function. Limited resources are in place at this stage to assess the newly installed tool.

This project takes post installation testing data and performs several statistical analyses to validate that the installed tool remains within spec post hardware installation. All data is collated into a report that remains with the tool for historical verification purposes and provides a first check to catch any issues as quickly as possible. This saves significant cycle time, having to go back to prior steps in bringing a tool online.

Python Training for Data Scientists and Analysts

Training for Data Scientists, Data Analysts, and other non-traditional programmer personas can be a challenge. Many lack formal training in software development and come from unrelated backgrounds. Python is the primary programming language for these personas. It is highly accessible and one can become productive very quickly. However, this does not mean one will be able to produce quality, maintainable, and scalable code. Oftentimes, the initial productivity hits a major wall as these developers contend with a mountain of technical debt.

We have put together many custom training courses to focus on these kinds of developers and the specific industry they are working in. The courses cater to the needs and abilities of the business organizations. These classes take a pragmatic approach to software development, focusing on the Python language, version control, IDEs, code quality, statistical analysis, data system design, and application development. We successfully help these non-traditional developers tackle their current technical debt, while preventing an ever-growing mountain of additional debt, so they can focus on solving problems and providing insight.

Evaluating ML Models on Sensitive Data

As machine learning has matured to provide critical insights in many areas, there is an ever-increasing need for proper data management to test these models. Improper management of sensitive data poses great business and regulatory risk. Many companies already have access control systems around their data, but no APIs available for their data scientists to respect these systems.

This project integrates the Microsoft graph APIs into a platform for testing machine learning algorithms to provide access controls. The end users can only see data and results for which they have proper access. With a combination of Microsoft Teams channel permissions and SSO, permissions are granted and removed easily and virtually instantaneously. This reduces the need for IT involvement to manage access controls, and gives teams working on specific algorithms tighter controls over their data and results.

Service Insight

When providing preventative maintenance services, it is important to provide your technicians with tools necessary to plan and validate their activities. They need to both understand if they are meeting their key deliverables (KPIs) to their customers, and if the machines they are servicing are developing problematic trends, outlier behaviors. Furthermore, particular customers require custom metrics to understand KPIs and machine behavior.

This project focuses on providing a platform for KPI insights with detection of trends and outliers, while providing a custom plugin architecture for custom metrics. This project reduces the planning burden for maintenance activities and helps technicians identify and resolve issues faster.

Democratized Access to Data

Many companies today have huge reservoirs of data from business processes, research labs, manufacturing, and various forms of monitoring. While the data is collected, many are unable to gain insight from this mountain of data. The first and largest obstacle is oftentimes access, not just authorization, but ability to query this data and mine it. It is a major problem when the domain experts within your organization don’t have the ability process the data collected.

Providing access to data via simple SDKs can go a long way toward providing business insight and generating enormous value. On multiple occasions, we have constructed client libraries in Python for non-traditional programmers to give them access to generate insight. Importantly, these libraries are not simply constructed to output data in a giant, generic dump. We focus on gathering domain understanding so that we can cater the library to the domain. Structuring data in a domain driven way accelerates the productivity of users. It democratizes access to data by making it accessible to users who might not understand how to structure data to solve their domain problems.

Systems Design and Review for Regulatory Processing

With limited resources or the need to meet regulatory guidelines, it can be difficult to develop systems that scale and lead to maintainable code bases. We have provide system design services to help meet these guidelines. In the past, we have reviewed existing code bases that manage complex systems and provided feedback to improve the designs.

When it comes to regulatory approvals, we must understand that our software is held to a high standard. Users of our end products have placed their trust, and in some cases their health and wellness, in our hands. We have not just a regulatory obligation to their safety, but an ethical one. We approach these designs and reviews from the standpoint of ethical obligation first to ensure secure, safe, and compliant software.

Decentralization for Data Sovereignty

Today, we are witnessing a concentration and amplification of centralized power over our data. Control of data and the insights it brings is a massive wealth generator, but it also poses serious privacy and security risks. These risks can have direct and sometimes detrimental impacts on the lives of our users. The current prevailing software architecture on the internet follows the client – server structure. While this has allowed many businesses to scale quickly, it has also introduced risk to business and their customers.

Can we do better?

It turns out that centralization isn’t the only way to produce scalable software. Recent developments in decentralized networking software are allowing us to build robust software that is local first (not requiring the cloud to provide functionality). Taking a pragmatic approach, we can leverage certain abilities of the centralized paradigm to bootstrap decentralized connections. We can have end-to-end encryption across these centralized points, ensuring privacy and security.

We are working on a platform that will allow your business data to be synchronized and stored in a decentralized fashion while giving you faster connections and data sovereignty. You own your data, and you should control how it is used.

Our MVP was inspired by a mom who asked if she had the right to make the decision to share pictures of her newborn daughter on her behalf on social media. Her daughter is not old enough to consent to having her personal life shared on social media, where it can be used in ways that verifiably remove any sense of privacy. This mom still wants to be able to share the precious moments of raising her daughter with close friends and family privately and easily. We want to empower this mother and daughter to share their moments while retaining sovereignty over them. Our MVP will provide a secure and private space to share messages and media with or without centralized points for backup and relay. Data is encrypted in such a way that these central points cannot read your data anyway! Stay tuned for more!