Hi, I'm Girma
📍 Ethiopia | Software Engineer @ Revvity Signals

LANGUAGES / TOOLS

Deep in builder mode learning how real systems scale by designing and breaking them.
I focus on backend orchestration, developer tooling, and understanding systems from first principles instead of treating frameworks as magic.

I enjoy projects where theory meets production: workflows, internal tools, and architectures that don’t collapse under real-world failure.

Luma -A RAG-powered agent using LangGraph, FastAPI, and AWS Bedrock (Llama 3.2).

OrderFlow – an event-driven order orchestration engine for multi-tenant commerce

STEM Ethiopia – an AI-powered adaptive learning platform for Ethiopian STEM students, built with Flutter and Appwrite, featuring personalized learning paths, skill assessments, gamification, and community-driven collaboration

Clicker – a user-driven, AI-enhanced Chrome extension for ethical web scraping, featuring Gemini AI-powered element detection, pattern recognition, table scraping, infinite scroll support, and workflow recording, with live previews and export options in JSON/CSV

PitchCraft – an AI-powered outreach tool that scrapes LinkedIn company profiles, stores raw data in Supabase, and generates personalized emails with optional AI analysis

Better_cursor – an AI-assisted coding environment exploring better feedback loops between editors and agents

Sora free watermark remover – A specialized extraction framework engineered to detect and eliminate embedded watermarks from OpenAI's Sora 2 AI-generated video content. Utilizing frequency domain analysis and pattern recognition, this tool achieves clinical removal without degrading visual fidelity.

• Safety-First AI: Guardrails and validation layers before LLM execution
• Asynchronous over Synchronous: Events over fragile API chains
• Failure as a first-class design input: Designing for agent hallucinations
• Observability before optimization: Tracking token usage and agent traces
• Documentation as part of the system

• Designing Secure AI Database Agents with restricted tool access
• Studying distributed systems by building them
• Creating tools I personally want to use
• Reading architecture docs more than tutorials
• Learning when not to add complexity

• Distributed systems
• AI Agent Orchestration
• Backend architecture & pgvector
• Developer experience (DX)
• AI as infrastructure, not hype
• Understanding how editors, CLIs, and backends really work

"Make the system observable, or it doesn’t really exist."

• Prefers logs over dashboards
• Enjoys tracing bugs across services
• Believes simplicity is usually discovered late
• Builds to understand, not to impress