available — open to staff & principal roles
Six years across the full stack — the distributed systems and internal platforms that let forty-engineer orgs ship, and the web and mobile apps that put them in users' hands. I optimise for boring reliability, sharp interfaces, and code review that leaves people better than it found them.
Rated by recency of use, not self-assigned skill bars. "daily" means it's open in a tab right now.
Metrics collection, dashboards, and alerting in a single static Go binary. Built to run on one box with systemd — no cluster required. Ships with Spring Boot actuator and Python/FastAPI examples, so wiring up a real service takes an afternoon, not a sprint.
An end-to-end pipeline: Kafka for events, Spark for batch, ClickHouse for analytics, Airflow for orchestration, MinIO for object storage, and a dashboard on top. The whole thing comes up with one docker-compose file.
Segmented, partitioned message queues with commit-log storage and replication for fault tolerance. Two storage engines — glommio for DMA-friendly I/O, in-memory for tests — plus a tokio server and a CLI client.
Terraform modules for a production-grade Kubernetes cluster on AWS: networking, ingress, WAF, service-account policy, and an observability stack with Grafana, Prometheus, and CloudWatch. Every concern is its own module, so each review stays small.
Ansible playbooks for hardening Linux systems to the CIS benchmark: converge, harden, lock accounts, and an OpenSCAP compliance scan. Tested across Debian and Kali through Molecule scenarios, with ansible-lint and yamllint baked into the workflow.
A React component library of accessible, customizable UI primitives — buttons, dialogs, forms, navigation — plus ready-to-use templates for landing pages, dashboards, and docs. Written in TypeScript, with accessibility checks wired into the test suite.
A cross-platform Flutter framework for collecting and streaming sensor data from iOS and Android with configurable sampling. One API over both platforms — no per-device plugin juggling. Built for mobile-first data capture that holds up in production.
The native-Android counterpart to Pulse: a modular, probe-based sensor framework with battery-aware sampling. For when you want the full platform API surface and tight control over the sampling loop, with no framework layer in between.
at Ethiopian Artificial Intelligence Institute ↗
Leading the platform team on the institute's engineering work — service frameworks, data infrastructure, and the tooling the rest of the org builds on. Addis Ababa.
A TypeScript analytics platform with a type-safe semantic layer for querying large datasets and building dashboards. The monorepo spans the ClickHouse connector, datasets, protocol, and an MCP server so agents can query the same layer the dashboards use.
A reusable Go framework for microservices and CLI tools — routing, gRPC, GraphQL, metrics, security, and the shared utilities a team keeps rewriting otherwise. Three example services show it end to end.
Fastest reply is email. Happy to talk staff and principal roles, consult on reliability and platform work, or just trade incident war stories. Replies within one business day — usually faster than the CI pipeline.
danielmekuriafeleke@gmail.com