PROBLEM · ARCHITECTURE · OUTCOME
Selected Projects
The kinds of systems I build. Proprietary work is intentionally summarized at an architectural level.
Cloud analysis and automation systems
Problem: Teams need reliable processing and feedback around engineering events and analysis workflows.
What I build: Event-driven systems using infrastructure as code, Lambda workloads, SNS topics and subscriptions, private networking, security groups, IAM policies, and API integrations.
Technology: AWS CDK, TypeScript, Python, Lambda, SNS, VPC, IAM. See the AWS work page →
Quality engineering platforms
Problem: Large product surfaces need consistent validation across services, web experiences, mobile experiences, and device-oriented workflows.
What I build: Automation frameworks, service-level validation, reusable test utilities, and tooling that turns failures into actionable engineering signals.
Technology: Python, TypeScript, JavaScript, GraphQL, Linux, Git, Docker.
AI agents and deterministic benchmarks
Problem: Engineering agents need to be useful, safe, and measurable. A convincing demo is not enough to establish whether an agent is improving.
What I build: Agents for test authoring, code and test migration, validation, and stabilization, supported by generated skills, reusable SOPs, and focused tool access.
Technology: Python, TypeScript, MCP tools, agent skills, benchmark orchestration. See the AI systems page →
Self-hosted automation
Problem: Useful household and personal workflows often span devices, network services, scheduled jobs, and external integrations.
What I build: Button-driven automation and operational tooling with explicit service boundaries, safe deployment, health checks, and recovery paths.
Technology: Python, Bash, Telegram interfaces, Linux, systemd, networking.