Brian de la Motte

I built a data model that unified Google Cloud Platform, Amazon Web Services, Packet, Digital Ocean cloud invoices into a star schema dimensional model.

I designed it from end to end and built all the data pipelines. This project was used internally to forecast the company's future cloud costs and to drill down into the specifics to see how money was being spent.

I architected, designed, and built a big data ecosystem nicknamed (BDP) for the United State's Social Security Administration. This was a Hortonworks stack on AWS 100% and automated by Ansible, CloudFormation, and Cloudbreak. The tech stack included Spark, Hadoop, Hive and more.

Out of this came an idea of dynamic templating so several clusters could be spun up in parallel and used for development or production.

The government had the most stringent requirements involving things being approved for FEDRAMPED and before using the cluster for official government use, we had to get the official "Authority to Operate" approval. This involved turning every security bell and whistle you could think of, including vormetric disk encryption, SSL on everything, security groups, firewalls, authorization, SSSD, Kerberos, RedHat IdM integration, cross-forest Active Directory trusts.

After getting the ATO, we helped several teams within the SSA get onboarded to the BDP to use Hive and Apache Spark for various ML and analytical workloads. Our work was so successful, the NIH wanted to use our cluster for some work too.

The ML work involved parameter tuning and installing TensorFlow on Spark, XGBoost, and other popular libraries.

I built a modern data lake on AWS for Time, Inc. This involved two Greenplum clusters and two Hadoop clusters and everything was automated with CloudFormation and Bash. Several clusters were also tied into the company's identity infrastructure.

I assisted in the migration of over 1PB from AWS to GCP. We migrated this data from S3 and Snowflake over to BigQuery and GCS. It took a total of ten engineers and three months. Dozens of data pipelines needed to be rewritten to work on Google's GCP Composer service (Google's hosted version of Airflow), tested, and deployed to production.

I migrated an entire web stack hosted in Azure on Cloud Services (Classic) over to a new Kubernetes cluster running on Azure AKS. This involved finishing up the migration from .NET to .NET Core and then getting .NET Core to build on Linux containers instead of on Windows successfully.

Finally, after getting the cluster pod autoscaling and horizontal pod autoscaling working correctly, I set up several CI /CD pipelines for production, staging, and development branches to automatically build and deploy the updated images to the production, staging, and development websites.

Out of all this, the company saved about $2,000 per month on their hosting costs as the autoscaling was much more resilient, and the new cluster was cheaper than paying for the raw VMs.

I built a modern web app for Empire Unmanned. This web app was responsible for allowing field pilot drones to upload their drone videos to the cloud to be processed for aerial drone inspections and mapping. In addition to being a back-end developer on the project, I was the engagement manager on the project and led a small team of engineers to complete the project.

I led a small team of software engineers to build out a small API written in Phoenix to replace a very old Perl-based system on its last legs. We got a basic API spec from the client and had to implement their needs while ensuring the API met the client's demands and was properly tested.

• Owned the audio capture and transcription platform at an AI meetings company, end-to-end, across Windows, macOS, iOS, and Android.
• Engineered the pipeline in Rust: multi-device capture, WebRTC AEC3 echo cancellation, resampling, and streaming ASR with fault classification, plus audio-in and words-out reconciliation that guarantees data integrity from microphone to database so a dropped word is detectable rather than invisible.
• Recovered 20% of captured audio that was being silently discarded by fixing resampling and downmix defects in the 44.1kHz path.
• Shipped on-device transcription on three platforms - iOS SpeechAnalyzer, Android Kroko/sherpa ASR, and a native macOS capture service, with a canonical-transcript policy that resolves conflicts when sources disagree.
• Designed the underlying PostgreSQL data model, ERDs, and architecture decision records, and the Goose migration pipeline that still runs every schema change.
• Took the product to market on a Tauri shell, then replaced it with native Swift, Kotlin, and Windows implementations, and launched both mobile apps through App Store and Google Play review.

I worked as the Fractional CTO for a property-technology startup building an NFC-tag asset and maintenance platform for multifamily apartment operators.
• Took over a stalled codebase and restarted it: the project had produced 22 commits in the four months before I arrived and shipped 902 in my first six months.
• Directed a team of six engineers delivering roughly 1,440 commits across three repositories in nine months. Consolidated three legacy applications into a single Next.js and TypeScript codebase, retiring a 228-commit React front end and two abandoned mobile apps.
• Put the production database schema under version control for the first time in the company's history with Prisma and a migration pipeline that ran on deploy.
• Directed a complete back end re-platform from AWS Lambda and MySQL onto NestJS, PostgreSQL, and ECS Fargate - 433 commits in 18 weeks, 23 controllers, 34 data models, 41 test suites, and a hand-written cutover migrating live production data.
• Shipped an AI document-understanding pipeline turning a phone photo of an appliance serial plate into structured manufacturer, model, and warranty records using AWS Textract and OpenAI.

I worked as Chief Technology Officer for a fitness-equipment maintenance startup, taking the product from nothing to a live multi-location customer in five months.
• Structured the engagement myself, trading a 25% discount on the build fee for a 10% equity stake, and delivered the MVP on a contractual hard deadline with the team demobilized on schedule.
• Directed a team of seven engineers and a project manager that shipped roughly 920 commits across three codebases.
• Designed the core product mechanic: a member scans a QR sticker on a machine with their phone camera - no app and no login, and the report becomes a tracked repair ticket tied to that machine's make, model, and warranty history.
• Oversaw a NestJS and PostgreSQL back end with true multi-tenant authorization, a Vue 3 operator console, preventive-maintenance scheduling, and SMS and email notification pipelines.

I now serve as a technical advisor and built the AI product surface solo: conversational equipment-fault diagnosis, where an operator describes a fault in plain language and receives a structured diagnosis, using OpenAI and the Vercel AI SDK on Next.js and React 19.