Ebrar Onuk

Full-Stack & AI Developer · Istanbul, Turkey

Hi, I'm Ebrar, a 26-year-old Electrical & Electronics engineer. I started building software through freelance projects about four years ago. Especially over the last year, I've been focused full-time on AI systems, agent orchestration, and RAG.

Right now I'm working on a multi-agent orchestration coordinating 30+ agents with LangGraph and LangChain: bringing together RAG, persistent memory, tool-calling, intent detection, and STT/TTS. Alongside that, I work on several React-based web projects.

In the same period, as one of two engineers on a small team, I took a multi-agent mobile application from idea to release on iOS and Android, reaching ~500 users.

Before that, for about a year, I worked on defense-industry projects at Baykar Tech and Pavelsis Avionics, taking part in configuration management, software tooling, and embedded software work.

Skills

Full-Stack Development:
ReactNext.jsTypeScriptJavaScriptReact NativeNode.jsExpressPythonREST API
AI / LLM:
LangGraphLangChainRAGAgent OrchestrationTool / Function CallingFastMCPIntent DetectionLLM APIsSTT / TTS
Data & Infrastructure:
PostgreSQLRedisSupabaseSQLDockerKubernetes
Tools & Approaches:
GitFigma MCPAgile / Scrum

Projects

Asisimo

AI family assistant, live on iOS & Android

Built from the idea stage as one of two engineers on the team. I owned it across every layer, from frontend and backend to AI/agent architecture, database schema, a shared component library, and UI testing, through the initial release and every update since. The system runs a multi-agent supervisor → orchestrator → specialist architecture built with LangGraph, with a 3-tier hybrid intent detection architecture and a multi-layer guardrail system.

  • End-to-end App Store Connect and Google Play Console ownership: builds, versioning, phased rollouts, and review submissions
  • Push notification infrastructure end to end (APNs/FCM)
  • Crash monitoring, release notes, and post-launch iteration
React NativeExpoTypeScriptNode.jsExpressLangGraphSupabaseRedisGoogle Gemini
App Store ↗Google Play ↗asisimo.com ↗

Multi-Agent Banking RAG

Open-source portfolio piece

Most of my professional multi-agent work sits in private company repos, so I built this one specifically to be public: a from-scratch reference implementation of the same architecture patterns (LangGraph orchestration, hybrid RAG with vector plus BM25, FastMCP tool-calling, rule-based NER/intent recognition, and a policy guardrail layer), wired around a fictional banking-support domain.

LangGraphFastMCPFastAPIChromaDockerKubernetes
GitHub ↗

nextreach-chatbot ↗

B2B SaaS lead-qualification chatbot and admin dashboard, built in a 6-hour timed challenge.

Next.jsSupabaseGemini

Previously

Only on GitHub