As a fully managed data labeling provider, Alegion works with various workforce contractors across the US and around the world. Over the past four years, our partnership with StepWise has driven huge growth. For StepWise, data annotation opportunities from Fortune 500 companies have helped employ and retain 180 Kenyan workers, many of whom would otherwise be unable to make fair living wages. For Fortune 500 companies, Alegion’s worker training and management, project management, and intuitive annotation software have produced millions of labeled data points quickly and cost effectively.
Who powers the global data labeling workforce? For much of it, business processing outsourcing companies (BPOs) do. Unseen by end customers but critical to the success of Fortune 500 initiatives, these sourcing companies train workers to help scale up IT-enablement projects at a moment’s request.
For the past four years, Alegion has partnered with StepWise to recruit highly skilled workers for large data annotation projects. StepWise is a BPO based in the United States and Kenya, and is known as an “impact sourcing provider” because it employs individuals from underserved Kenyan communities. In fact, StepWise is a certified B Corp, which, according to social enterprise Close the Gap Kenya, means it “meets the highest verified standards of social and environmental performance, transparency, and accountability." Vice President of Marketing Sarah Wakasa says that StepWise is “equally passionate about purpose and profits.” Together with Alegion, StepWise has economically empowered 559 women (50% of its workforce) and 67 people with disabilities (6% of its workforce) in 2021.
"Our strategic partnership with Alegion provides our BPO workforce with a powerful data labeling solution platform to help manage the most complex labeling needs with efficiency, accuracy and agility,” said Alex Gichuru, Chief Operating Officer of StepWise. “Our accredited workforce gets access to a first-class platform to deliver quality services to our client base globally."
Using a Sourcing Partner to Scale Up ML Initiatives
Over the years, Alegion has partnered with StepWise on a range of projects. These include training a model to identify potential threats in public spaces to sentiment analysis using optical character recognition (OCR) on hospitality reviews. Oftentimes, the Fortune 500 companies that are outsourcing the work to Alegion and its partners require thousands of labeled data points to train their models to a 90% or greater accuracy rate.
For one big box security data labeling project, Alegion communicated to StepWise the customer’s timeline, the volume of work, and project guidelines. StepWise then tapped into its network of 380 skilled annotation workers and, within two days, identified and qualified a pool of 100 workers. Each worker was qualified with a strict labeling test based on the customer’s gold dust data. Qualified workers then began accepting batches of work according to workflows structured by Alegion’s data experts.
The security project required three million data labels in just four months, so StepWise workers produced 25,000 labels per day and saved the client 960 hours’ worth of data labeling. And by using trained workers to identify human figures that were up to 200 feet away from the camera, the StepWise and Alegion partnership produced training data that could refine the machine learning model to make event predictions much sooner than the average model, which might wait until a human is within 50 feet before delivering confident predictions.
Training Workers on Alegion’s Software
The customer provided thousands of generated videos for data labeling. This image is an example of bounding boxes drawn around individual grocery items associated with a shopper.
this video frame provides an example of people annotated with bounding boxes The current workflow asks for all people within a range of 100 feet to be labeled.
Using Alegion’s software, workers can use bounding boxes, polygons, key points, and semantic segmentation to localize all required objects in a given video or image. In the big box security project, workers annotated thousands of videos that lasted one minute each. Because of the efficiency of Alegion’s platform, they reduced annotations to every 7th frame because of Alegion’s automated labeling through interpolation.
Victor Otieno, the project manager for StepWise’s computer vision and natural language processing divisions, shared workers’ experience using Alegion’s platform over others: “The nature of the tasks exactly matched the capabilities of the platform. Based on the instructions we received to complete the task and the way the workflow was set up, we could complete the work smoothly and efficiently. We haven’t had a single issue reported with Alegion’s platform.”
An Ongoing Partnership to Create Reliable Training Data and Fair Jobs
The quick turnaround, high throughput, and high accuracy of StepWise’s annotation workforce makes them a formidable partner in the age of machine learning. Wakasa describes StepWise and Alegion’s partnership as a perfect marriage of services: “Alegion has the power of technology and we have the pre-trained workforce so the client doesn’t have to worry about delivering the solution.”
Through a partnership with business processing outsourcing provider StepWise, Alegion creates jobs to support local Kenyan economies and helps fight social inequalities by hiring women, young professionals, and people with disabilities. For Fortune 500 companies, Alegion’s employment of these skilled IT workers through StepWise generates enormous amounts of training data to get machine learning projects off the ground.
Alex Gichuru, Chief Operating Officer of StepWise, shares his hopes for the future of machine learning. “AI/ML presents a unique opportunity to become the single largest engine for job creation for underserved communities, particularly women, youth, and persons with disabilities,” he says. “Working together in this partnership sustains our purpose of putting Africa at the forefront of the Fourth Industrial Revolution."
You can learn more about the change we strive to make through our Impact Centers. If you’re looking to get high quality training data faster and at scale, work with Alegion and its partners. Speak to a machine learning expert today.