# Topic Backlog — AI × Networks Research KB

Prioritized by intersection research value. Update this as you complete topics and discover new ones.

## Priority 1: AI × Networks (Intersection Domain)

These topics form the core research edge. Start here.

- [ ] ML for RAN anomaly detection
- [ ] RL for network slicing and resource allocation
- [ ] Inference at the edge (constraints, architecture)
- [ ] LLM for network configuration generation
- [ ] Transformer models for traffic prediction
- [ ] Federated learning in RAN environments
- [ ] Deep RL for RAN optimization
- [ ] Graph neural networks for network topology learning
- [ ] Reinforcement learning for load balancing
- [ ] Neural architecture search for edge devices

## Priority 2: Networks (Current Work Domain)

Pure networking concepts that ground intersection work.

- [ ] ORAN architecture (CU/DU/RU split, xApp lifecycle)
- [ ] 5G NR numerology and frame structure
- [ ] Network slicing + SLA guarantees
- [ ] RAN scheduling algorithms (proportional fair, round robin)
- [ ] Fronthaul constraints and latency budgets
- [ ] Beamforming and massive MIMO
- [ ] Cell-free RAN architecture
- [ ] Network function virtualization (NFV)
- [ ] Software-defined networking (SDN)
- [ ] Network traffic modeling and prediction
- [ ] QoS and traffic engineering
- [ ] Network security and anomaly detection

## Priority 3: AI (Learning Direction)

Pure AI/ML concepts for building ML engineering skills.

- [ ] Transformer attention internals (mathematical derivation)
- [ ] KV cache mechanics and inference memory layout
- [ ] RL policy gradients (REINFORCE → PPO)
- [ ] LLM serving on K8s (latency/throughput tradeoffs)
- [ ] Diffusion models (score matching, DDPM)
- [ ] Graph neural networks (message passing framework)
- [ ] Quantization and pruning for edge inference
- [ ] Model compression techniques
- [ ] Attention mechanisms beyond transformers
- [ ] Sparse neural networks
- [ ] Continual learning and catastrophic forgetting
- [ ] Multi-task learning frameworks

## Priority 4: Papers

Research paper analyses that feed into above domains. Add as you discover relevant papers.

- [ ] [Paper Title] — [citation] — maps to: [networks/ai/ai-x-networks topics]

---

## Completed Topics

(None yet — this section will grow as you finish topics)

---

## How to Add Topics

When you discover a new relevant paper or concept:

1. Identify its primary domain: Networks, AI, AI×Networks, or Papers
2. Add to the appropriate section above
3. Note if it blocks or connects to other topics
4. When starting, move to "In Progress" and track via tasks
5. On completion, move to "Completed Topics" with link to the doc

## Status Legend

- `[ ]` Pending (not started)
- `[/]` In progress
- `[x]` Completed