Situation
The client, a global insurance company, came to Sketch when it was ready to take a bigger step into the cloud. They knew they needed a partner with expertise in cloud development and cloud optimization. They also wanted to partner with people who had deep experience in the insurance industry. Insurance companies must remain aware of specific regulations and requirements, so Sketch’s background was ideal for this client.
Because this client was still new to AWS Cloud and serverless technologies, they needed a partner who could walk alongside them and lead by example. Training alone wasn’t going to be enough, especially because this client had low risk tolerance and limited learning time. The engagement would have to involve a high degree of collaboration and co-development.
Long run times to process large amounts of data
The massive scope of this client’s operations created a second challenge. Because the client was processing large amounts of data, run times were becoming unacceptably long. The client needed a way to reduce or eliminate these problematic delays.
The client needed efficient, scalable software to reduce costs.
Need to scale legacy software?
We'll tell you how we did it (for free).
Serverless deployments and modern programming languages
Sketch’s lead developer created a platform to facilitate best code practices for serverless deployments using a wide variety of AWS resources, including Lambdas and ECS Tasks.
Next, we delivered a serverless template repo to jumpstart other developers quickly. From there, creating and delivering a React and Node.js workshop equipped the client’s developers with the language skills they needed.
With the team sufficiently upskilled, we did the following to equip them for success by improving the tools at their disposal:
- Enhanced a Node.js parquet library merging features from multiple sources so the client would have the most comprehensive and complete version
- Built a distributed logging and metrics library shared amongst the company
- Built a functional dependency injection system that would facilitate software development with Node.js and AWS lambdas. This simplified the code and made it easy to test in isolation
- Designed a scalable, distributed financial projection export clustering system built on EC2
- Designed and built an extremely scalable cloud load ETL platform based on lambdas, step functions, dynamo DB, and S3
- Enhanced the MS SQL Node.js driver to give it the first-class Node.js streaming capabilities it was previously lacking
- Created an ETL system that leveraged lambdas, step functions, dynamo DB to stream between MS SQL Server and Oracle
- Refined Infrastructure as Code (IaC) capabilities using Terraform
- Introduced a Rust-powered tool to efficiently convert parquet files to CSV for bulk loading into databases, and added a feature to the tool to support the client’s use case
- Rapidly designed and developed an elegant bulletproof Database load system using ECS tasks to replace a problematic Enterprise Platform that was continually overloading the Database, causing wide outages — leveraging the Rust parquet to CSV tool and a few other Rust tools for partitioning for the heavy lifting of extremely efficient conversion and loading of very large data sets
- Delivered an Okta-protected micro-UI platform using the AWS CDN and S3 for hosting several custom applications
The remaining effort largely came down to providing leadership in the form of training and mentoring. We helped the client establish best practices and adopt new techniques. Everyone on the team became comfortable with what Sketch had delivered and co-developed with them. Video training resources ensure future developers can come up to speed quickly, too.
Let's build better software together
If these results (saving thousands of dollars per run, developing new software systems when you need them, and upskilling your internal software development team) sound like what you're looking for, let's talk more.