Andrzej Wodecki: Thesis Topics and Resources
Potential Thesis Topics
(Multi-)agent LLM-based Systems
- Emerging behaviors in L/SLM-based multi-agent systems
- Communication
- Architecture
- Multi-agent systems based on SLMs deployed on edge devices (core topic of my research)
- Emergent behaviors
- Deployment and monitoring
- Applications, specifically in industry 4.0
- SLM optimization (pruning, tuning, etc.) for multi-agent applications
- Agentic MESH
- Agents
- Registry
- Discovery
- Execution
- Experience planes
- Evaluation and monitoring of multi-agent architectures
Optimization of LLM Systems
- LLM design and optimization methods based on DSPy
- Self-learning LLM systems (source)
LLMOps: S/LLMs in Production Environments
- Testing the effectiveness of different strategies for LLM deployments (source)
Knowledge Resources
Foundations
Embeddings
Transformers
Pre-training and Fine-tuning
Optimization
Specific Models
Agentic AI
Agents
Multi-Agent Systems
Applications
Deployment
RAG (Retrieval-Augmented Generation)
Data Sources
Vector and Graph DBs
Agentic RAG
Evaluation
Optimization
Multimodal
Applications
Prompt Engineering
Applications
Software Engineering
Frontend
Business
Deployment
Backend
Frontend
Edge
LLMOps
Evaluation
Models
RAG
Agents
Applications
Multimodal LLMs
Safety and Reliability
Last updated: March 2025 | AI Thesis Topics and Learning Resources