Head of AI & Founding Engineer | Multi-agent systems | LLM infrastructure | Production GenAI
Leading AI engineering at Wubble AI - real-time multimodal audio generation (music, vocals, SFX) and agentic workflows at scale.
Agentic & LLM systems
- LangGraph multi-agent systems (HITL, routing, tool-calling)
- RAG and LLM deployment across major model providers
- MCP dev tooling (langmcp)
Multimodal & real-time AI
- Real-time audio generation (music, vocals, SFX)
- Voice AI (streaming STT -> LLM -> TTS)
- Transformer serving on Kubernetes
Production infrastructure
- Multi-cloud GPU stacks (AWS/GCP) with routing and autoscaling
- Event-driven microservices (FastAPI, Pub/Sub, Redis, PostgreSQL)
- MLOps: CI/CD, IaC, versioning, monitoring
Applied ML & vision
- Computer vision and real-time detection (e.g. YOLOv8)
- OCR and document pipelines
- NLP/CV research (captioning, generative models)
LoRA fine-tuning | instruction tuning | RAG | prompt engineering | model optimization
Multi-agent orchestration | HITL workflows | tool-calling | agentic AI | semantic search
AWS: EC2 | Lambda | S3 | SageMaker
GCP: Cloud Run | Cloud Functions | Pub/Sub | GKE | Vertex AI | Cloud Storage
Infrastructure-as-code | model versioning
B.Sc. Electrical and Electronics Engineering, Bilkent University (Honors, Cum Laude).
Previously: LAYMARK (multi-agent and Voice AI) | Simply Complex Lab @ National Nanotechnology Research Center (deep learning for microscopy) | applied ML across CV, OCR, and generative modeling.
- LinkedIn - linkedin.com/in/m-abdullah-mulkana
- Medium - medium.com/@muhammad.shafat
- Substack - muhammadasmulkana.substack.com
- GitHub - you're already here


