Streamlining Data Analysis Pipeline

Industries: Manufacturing, Robotics, Transportation
Services: Big Data & Analytic Services, Custom Software Development & Engineering, Quality Assurance & Testing

The Company: An Accessibility-Focused Tech Company

A tech company assisting people with disabilities in navigating cities around the world approached us about streamlining their internal processes. Co-founded by an individual with physical disabilities, this company aspires to become the Google Maps of pedestrian pathways, leveraging numerous data points to provide handicap-accessible routes that are free of dangerous obstructions. Using a mobile telemetry device, the company scans sidewalks and curbs to detect tripping hazards and a variety of other accessibility issues. To date, our client has recorded more than 3,000 miles of sidewalks and identified more than 50,000 hazards and obstructions, all to help build a more ADA-compliant world.

The Challenge: Automating A Manual Data Analysis Process

After years of relying heavily on a research-focused program for a manual and labor-intensive data analysis process, the client’s team found themselves struggling to keep up with increased service demands from their customers. As a result, the company aimed to automate as much of its data analysis as possible while simultaneously reaching three major goals:

  • Port the existing algorithm to a different programming language.
  • Break their dependency on their current program licensing.
  • Avoid another major investment in licensing.

The Solution: Develop New Code to Streamline Data Analysis

Our team started by identifying a suitable tech stack to meet the client’s unique data analysis requirements. From there, we identified a compatible workflow automation tool (Dagster) to further streamline their processes. Then, using Python, we converted their code file-by-file, constantly checking for accuracy and fidelity. Through this process, we identified and removed numerous bugs and then optimized the code to more than double its throughput in the same amount of time. After that, we ensured the code was set to run automatically whenever data points were recorded. Finally, we built a new dashboard in React, which pulled data from Dagster using GraphQL and ArcGIS via REST API. This dashboard provided a display for the company to review data with their clients.

The end result: Our client is now ready to scale at a much faster pace while delivering better insights to their customers.