I was bitten by backend engineering somewhere around a computer science degree with an AI emphasis at the University of Georgia, and I never quite recovered. Most of my work now lives at the intersection of the two. Systems that are fast and reliable underneath, with an ML aware layer swinging on top.

The unglamorous parts of the job are half the appeal. Chasing a flaky service through a stack trace at midnight, writing a Terraform module boring enough that nobody has to think about it again, catching a timezone bug before it catches me. If a system needs to be observable, reproducible, and operable by someone who is not me, that is the kind of web I want to build.

Certifications earned along the way: AWS Solutions Architect Associate, AWS Cloud Practitioner, and Azure AZ-900.

Every hero has an origin experiment. Mine is InferGrid, a production grade distributed ML inference platform built on GCP. It is the project where backend engineering and applied AI actually fuse into one working system instead of living in separate corners of a portfolio.

• Kubernetes based autoscaling for inference workloads
• A/B model routing between DistilBERT and a scikit learn baseline
• Kafka backed async request queue
• Drift detection on live traffic
• Sub 100ms p95 latency at 500 plus requests per minute

A live demo dashboard, built in React and deployed on Vercel, talks to the GCP backend in real time so the whole thing is watchable, not just describable.

AtlasFlow. An event driven backend built on DynamoDB and SQS, tuned for decoupled, asynchronous processing rather than one long synchronous chain.

Nexus. A distributed observability platform for tracing requests and monitoring service health across a cluster, the kind of tool you build so the next 2am page is shorter.

Procuroid. A multi agent procurement system that earned an Honorable Mention at AI ATL 2025 on the Google Cloud track.

FraudSense. Fraud detection using graph structured models, built to catch the patterns that flat tabular data quietly misses.

PixelVault. A zero cost photo archival pipeline that routes iPhone photos through a Vercel relay to an old Pixel XL, unlocking lifetime unlimited backup through Google Photos. Built with a Next.js backend, a Kotlin Android app, and iOS Shortcuts on the upload side. Less of an AI showcase, more of a fun systems automation heist.

Mini ML Platform. A smaller scale run at the ideas behind InferGrid, the proving ground before the real thing got built.

Languages

Backend, Cloud, and Infra

Data, Messaging, and Observability

With great scale comes great responsibility, for uptime, for logs, and for whoever is on call when things break at 2am.

Build systems that are observable.

Build systems that fail gracefully instead of embarrassingly.

Build systems someone else can swing through without needing a map.