Cloud infrastructure · distributed systems · developer tooling

I'm a senior engineer at AWS, a maintainer of OpenSearch Migrations, and a contributor to Floci, an open-source AWS emulator. I build concurrent Java systems, cloud API emulation, Kubernetes deployment tooling, and integration tests.

183 merged upstream pull requests, spanning workflow execution, service compatibility, and concurrency.

• Workflow and runtime execution. Implemented SWF emulation covering 39 modeled operations, including workflow state transitions and activity lifecycles. Added Lambda provisioned-concurrency runtime initialization, routing, and cleanup.
• Concurrency primitives. Built bounded work queues and versioned signals to limit workers and retained keys, coalesce notifications, and prevent concurrent execution for the same key.
• Compatibility verified against AWS. Corrected Step Functions callback authorization using comparisons against real AWS APIs and signed SDK regression tests, including resource-scoped grants, explicit denies, and completed-token retries.

Maintainer · 515 merged pull requests · Maintainer roster

• Java memory and concurrency. Replaced a large in-heap term-position index with external sorting and memory-mapped sidecars for source reconstruction. Fixed a Netty event-loop deadlock by moving blocking response collection off the event loop.
• Kubernetes deployment and CI. Parallelized EKS provisioning and teardown and extracted shared Jenkins functions, improved bootstrap readiness and cleanup after partial failures, and contributed GitHub Actions test sharding.
• Incremental migration and correctness. Authored the delta-snapshot RFC and implemented update and delete processing using Lucene segments and live-document bitsets. More recently, corrected bulk request limits using exact serialized sizes.

I maintain docker-cache, a fork of ScribeMD/docker-cache for caching built and pulled Docker images. Released 0.7.1, fixing cache lookup consistency across parallel jobs after updating the runtime and cache dependencies in 0.7.0.

Earlier versions appear in public CI workflows in OpenTelemetry Collector Contrib, LACT, and FS Crawler.

• snap-version — inspect and edit version metadata in offline OpenSearch and Elasticsearch snapshot copies, with format validation and checksum handling. Metadata edits do not convert index data or establish restore compatibility.
• Jira MCP Skill — configure selected coding-agent clients to launch the upstream Jira/Confluence MCP server while preserving other client settings. Published as @andrekurait/jira-mcp-skill on npm.
• OpenSearch Migrations demo — a Rust TUI for planning and provisioning disposable migration test environments, with a review step before provisioning.
• AWS agent toolkit — contributed the Amazon OpenSearch Service skill, covering migration, provisioning, search, and observability guidance for coding agents.

• acp-traces — an experimental Rust proxy that adds OpenTelemetry tracing to Agent Client Protocol interactions, including prompt turns, tool calls, and timing.
• TemporalCI — a historical proof of concept for CI build orchestration with Temporal workflows and Kubernetes test-cluster lifecycles. Preserved as a reference; no active platform or deployment is maintained.

Contribution counts checked November 2, 2026; all-time upstream merged PRs only.

LinkedIn · Email