LIMS-to-platform integration
The problem
A sample-processing workflow at a genomics lab was running almost entirely by hand: someone would pull data off scanned documents, retype it into a spreadsheet, run two separate scripts on a server, update the spreadsheet again, then manually update the LIMS. Each sample took two to five minutes of a team member's time, sequentially, one at a time. It was slow, it ate hours from a team that had better things to do, and worse, it had already let real errors through more than once.
What I did
Rather than build new infrastructure from scratch, I plugged into a DynamoDB table the team already had feeding their LIMS integration, reuse over reinvention. On top of that, I built an event-driven pipeline: DynamoDB Streams trigger a validation Lambda the moment new data lands, and an SNS topic with filtered subscriptions splits the result into two paths, a failure path that posts a detailed alert straight to the team's Teams channel, and a success path that produces the files needed to auto-update the LIMS and the sample-tracking platform.
What came of it
Once data enters the pipeline, samples now process in parallel in seconds, down from two to five minutes each, done one at a time, by hand. Manual work per sample dropped to about a minute of data collection; everything after that is automatic. When I ran a validation comparing the automated results against the old manual process, the automation caught errors the manual process had been missing, showing that the new system can be more accurate.
Who built this
I started on an operational bioinformatics team, the person you called when something broke, solve it fast, code it fast, keep it simple. Our platform ran on AWS, and a nearby team, Bio DevOps, was building it properly: real infrastructure, not patches. I got close to them, studied on my own, started writing small automations in boto3 and Lambda. Within a few months I was building in CDK, deploying real workloads, and the team's coordinator brought me on. Today I work as a (Bio)DevOps developer, building the automation and integrations that connect cloud infrastructure, LIMS, and bioinformatics pipelines, and I understand cloud architecture well enough to own it end to end, not just operate inside someone else's design.
Get in touch
If this looks like the kind of problem you need solved, send a message through the form on the home page, or reply to the email below.