As an emerging healthcare startup grew, they turned to PromptWorks for the expert developers they needed to build their healthcare recommendation engine better and faster.
Picwell – a fast growing healthcare startup – had a brilliant idea: to build a data-driven recommendation engine for healthcare plans. Using billions of data points from anonymized claims, they could predict which plans would save consumers the most money.
Your work has consistently been stellar and we couldn’t be out there helping hundreds of thousands of people if it weren’t for our partnership… It means a lot to have you guys along with us for the ride.–Ani Vemprala, CTO & Co-founder at Picwell
But like many emerging startups, Picwell had a bottleneck: They didn’t have the all the developers they needed in-house to build the complex data pipeline and robust API the product required.
Furthermore, as Picwell added new clients, they needed to build new models, often on extremely tight, client-driven deadlines. They needed the ability to boost their development capacity on demand to meet these deadlines.
Many fast-growing startups face similar situations. Here’s what we’ve heard from these startups:
- “We’ve got the idea, we’ve got funding. Finding the right talent is what’s holding us back now.”
- “We’re facing tight deadlines, but we also need to build this the right way.”
- “We don’t have the capacity to do everything in house, but we need to maintain ownership of our product throughout the process.”
By working with PromptWorks, Picwell and other startups are able to overcome their bottleneck in development talent, getting the talent they need but with flexibility for future growth.
By using PromptWorks, Picwell instantly had a dedicated team of expert developers, working seamlessly with their own team.
Picwell needed something more than just “developers for hire.” They needed developers with deep expertise in complex topics, like statistical modeling, automated data pipelines and robust API development. And they needed true partners – a seamless team that was in it for Picwell’s long-term success.
So we put together a dedicated team of experts, built around the needs of Picwell.
This provided a number of advantages to Picwell:
As Picwell faced client-driven deadlines, they needed to boost up the amount of support they received. But over time, as Picwell grew its own team, they needed the flexibility to scale down the amount of support they received.
Picwell needed more than just developers. They needed developers who were experts at complex statistical modeling, managing multiple data pipelines and building best-in-class APIs.
To hit its aggressive deadlines and ensure the product was ready for open enrollment, Picwell looked to PromptWorks for our expertise in Agile consulting.
Because Picwell’s main service is the API it offers its customers, it was absolutely essential to have quality assurance built into the product, with a suite of end-to-end integration tests and extensive monitoring of issues.
Throughout the process, Picwell needed to maintain full ownership of their own software, something that PromptWorks provides with no contract lock-ins and the ability to transfer the entire codebase at any time.
By understanding the needs of Picwell, PromptWorks was able to put together a team of expert developers and Agile consultants that could work seamlessly with Picwell to build a successful product.
Working as one team, PromptWorks helped Picwell build a better healthcare recommendation engine, on-time & on-budget.
Once in place, the integrated PromptWorks team focused on the three main areas of Picwell’s core offering.
First, PromptWorks helped Picwell improve their data model.
Picwell relies on multiple data streams to create their statistical models and to adapt to changes in health care plan structure. We helped them build a custom, fully-automated data pipeline using Luigi and Jenkins, which does the following:
- Ingests data from multiple sources (claims records, health care plans, formulary data)
- Extracts the relevant data from the source, cleans it into an ingestible format, verifies that it is not malformed nor has missing data, transforms it into a standard, common format, and stores it internally
- Runs simulations of healthcare costs across a distributed compute cluster
- Trains statistical models with the resulting simulation
- Performs quality assurance on the model to ensure accuracy and efficacy
- Deploys the model to production
(To build the statistical models, PromptWorks used its expertise in the Python data science stack, including Numpy, Scipy and Scikit-Learn).
Second, PromptWorks helped Picwell build a robust API.
Picwell’s clients depend on their API to provide timely and accurate responses. This is a core part of they value they provide clients, so it is absolutely necessary for the API to be highly available and highly accurate. Using Python and Flask, PromptWorks helped build a robust API.
To ensure the API is meeting client needs, it is supported by:
- Regular deployments of new features
- An exhaustive suite of end-to-end integration tests
- Extensive monitoring of issues and quick response to bugfixes
And third, PromptWorks helped ensure a smooth, successful project using our Agile expertise.
Our Agile process helped to ensure the right software was being built, and that it would be delivered on-time and on-budget. With our Agile development process, Picwell was able to see progress every day and could prioritize features in real time, so there were no black boxes or last minute surprises.
As a result, Picwell met their aggressive deadlines and their healthcare recommendation engine saw better reliability, better accuracy and quicker turnaround for building new models.
PromptWorks’ approach to team augmentation allowed Picwell to maintain full ownership of their software, but still receive the support they needed and meet their deadlines.
Here are some of the results Picwell has seen on this project:
- Reduced time spent building new models. With the help of PromptWorks, Picwell was able to bring down the time to build a model. What used to take weeks now takes just days.
- Increased confidence in model accuracy. By adding automated testing and quality analysis around the model creation process, we were able to increase the confidence in the end result, which allowed us to narrow the timeline even more.
- Highly Reliable API. By maintaining a suite of integration tests, as well as load testing the entire infrastructure, we ensure that the API is highly available, especially during peak traffic.
Since we began working with Picwell, we have continued augmenting their team to various degrees over the last two years, providing developers, technical leadership and expertise, as well as agile and business consulting. As Picwell has been able to hire more internal staff to fill the need, and become more efficient, we have been able to downsize our support accordingly.
We love helping startups like Picwell reach their full potential faster. So if you need expert developers to give your startup a boost, we’d love to chat about your needs, and help craft the perfect team for you.