- 🔭 Day job: Full Stack Developer & AI Engineer at WAGMI.Tech
- 🤖 Building: Production agentic systems with LangGraph, AWS Bedrock AgentCore, and self-hosted ML
- ✍️ Writing: Bilingual (TR/EN) AI engineering content on LinkedIn | X @yusufErdoganAI | Substack
- 🌱 Exploring: LLMOps, agent evaluation, cost-efficient local inference
Production document summarization pipeline built from scratch — infra, model deployment, orchestration, and evaluation
🔧 Stack: Python, LangGraph, AWS SageMaker, Bedrock AgentCore, ChromaDB, Lambda
📈 Results: +34.8% factuality improvement (QAGS: 0.354 → 0.477) over single-agent baseline across 50-query benchmark
Repos: multi-agent-summarization | multisum-agent-agentcore
Multi-platform AI agent (Telegram, Discord, CLI) with tool calling, Mem0 memory, and per-request cost observability
🔧 Stack: LangGraph, Python, Mem0, Claude
💸 Impact: Reduced LLM costs by 40% through granular token tracking daemon
Published skills for the developer community
- clawdbot-glm-ocr-local — Zero-cost local OCR via 0.9B GLM model (no API fees, full privacy)
- openclaw-qwen-tts-skill — Voice generation + Remotion video render
- Personal-Brand-Video-Skill — React-based video production via Claude Code
💡 "Ship production AI systems. Measure everything. Cut costs ruthlessly."