Infrastructure & migration
Production AWS designed for the load you actually have: ECS Fargate, RDS, load balancing across zones, DNS and certificates. On-prem servers and monoliths moved onto it without downtime.
Whole systems, one engineer.
I design, build and run the whole stack: AWS infrastructure, delivery pipelines, the application, and the AI inside it. Then I hand it over clean.
- 15 years in production
- AWS Certified: Cloud & AI
- DC metro & remote
What I build
Most teams hire four people for these and spend the project reconciling them. I do all four, so the joins between them are already done.
CI/CD & DevOps automation
Pipelines that take code from commit to QA, staging and production the same way every time. Containers, environment parity, and deploys anyone on the team can run.
Full-stack SaaS builds
Python applications with FastAPI, Django or Flask, payments through Stripe, bank connections through Plaid, and credentials encrypted from day one.
AI agents & LLM integration
Agents that do real work with an audit trail: classify requests, run routine operations, stream answers to users. Built on Claude and OpenAI, designed to degrade gracefully.
Selected work
Three engagements, end to end. Every claim below traces to work that shipped.
Smata Systems
A personal-finance and small-business accounting platform with nine integrated tools, live with paying customers. I architected, built and deployed all of it.
Visit smatasystems.com- ECS Fargate behind a load balancer across six availability zones, with RDS MySQL, Route 53 and ACM
- Plaid OAuth for live bank connections; Stripe Checkout for billing
- OpenAI and Claude insights, streamed in real time
- A 7,400-line monolith refactored into modules without downtime
An autonomous operations agent
An agent that watches incoming engineering requests, classifies them, and completes the routine ones itself: data loads, deployments, configuration patches. Every action leaves an audit trail.
- A multi-hour manual sequence became a single run
- Python and FastAPI across three model providers
- Monitoring dashboards that show what it did and why
From brainstorm to buildable
The founders arrived with an unstructured product brainstorm. They left with a system map, an architecture, and a platform decision made on stated constraints.
- System map and architecture from structured discovery
- A stuck platform decision resolved, rejected options recorded
- AI personalization designed in three stages that degrade gracefully
Two ways to work together
Engagements are contracted through Smata Systems LLC, remote or in the DC metro area.
Technical advisory
For deciding what to build.
Structured discovery, a system map, architecture and platform decisions, an AI assessment, and decision records you keep. Due diligence on a codebase or vendor, too.
Build & ship
For getting it built.
Product development, AI features, financial-data integration, cloud infrastructure, automation and security, shipped in increments you can see.
- Every recommendation tied to a named constraint
- Rejected options written down
- Security from the first commit
- Honest risk, said early
- One accountable owner
- Every engagement ends with a handoff
Experience
- 2025 – now
Founder & Principal Engineer
Smata Systems LLC · Mitchellville, MD
- 2014 – now
Senior Software Engineer
Federal agency · Maryland
- 2015 – 2020
Systems Architect
SAUCE Concierge Services · Mitchellville, MD
- 2011 – 2014
Systems Software Engineer
Space Telescope Science Institute · Baltimore, MD. Data pipeline tools for the James Webb Space Telescope.
Start a conversation
Tell me what you're building and where it's stuck. I reply within one business day.
Prefer email? info@robdavisjr.com