Pacing is specialized in custom application development. The Pacing methodology is based on full stack developers. Every developer has knowledge and understanding of all aspects of building applications, from the front-end back to databases and other data sources.
With that said, every developer also has one or more focus areas where they excel, for example: advanced search, big data or data science. In the development process we utilize every developer’s particular skills to construct systems that are efficient and reliable.
Custom applications give companies powerful competitive advantages. Most often there are business needs that cannot be achieved through traditional ERP or other systems of record. By combining powerful technologies like advanced search, machine learning and big data with existing systems you can realize customized user experiences and tailored processes that achieve competitive advantages.
Most data around us does not have a natural schema. This is the reason why schema-less or NoSQL data stores have emerged.
You apply a schema to the data in your store when you need to. Schema-less data stores like MongoDB can store extremely complex data structures that can change and evolve over time, without the restrictions of huge relational models.
You can store entire business objects like customers, orders or invoices and combine and retrieve them simply by fetching JSON-documents. We use MongoDB as our universal data store. Here we can dump raw data from just about any data source without paying much attention to how it is structured. The power of MongoDB can be used in many ways. We have cut the development time of applications involving complex data to a third and moreover some of the applications we have developed would not have been feasible using relational technology.
Search is not only about retrieving web pages. Search can be turned into applications – search on steroids. One example is where you have a textual requirement of an entity and want to find potential candidate matches. In practice somebody wants to buy an item with a set of requirements and you want to match this with your best fitting product. We have used Elasticsearch to perform this type of fuzzy matching in a number of applications successfully. Fuzzy search is also a great tool when you want to do initial data mining in huge data lakes.
Advanced analytics or machine learning in one form or the other is at the hart of most innovative applications. Today the problem is not lack of data, on the contrary we are living in an overcrowded data lake.
Any data driven project involves a lot of experimentation involving mining, preparation and visualization to explore what patterns or correlations may be found in the lake. But, the end goal should always be to deliver a working application like any other software project.
Being a software company we concentrate on practical applications of statistical methods to solve most every day problems. Our data science toolbox is selected for that reason and include RStudio Server, Base R and the Tidyverse package combined with a few additional specialized packages. We have developed our own production grade execution environment for R which provides monitoring and multi threading. It handles thousands of billions calculations every day of the year and provides vital business insight to hundreds of users.