Problems / LLM serving, easy
A fine-tuned candidate model is deployed as an adapter on the production GPU pool, and a traffic mirror copies 100% of live requests to it to compare answers. Production is 60 requests a second on a pool that serves about 80.
Unlock it once for 10 credits; everything inside it is free (runs, changes, Submit, results, hints). Sign in with Google to start; new accounts get free credits.
Success ≥ 99.5%, p99 ≤ 5 s.
Like a coding problem, a solution has to pass every case, not just the one it was built for. Submit in the lab runs all 4; 1 of them are hidden until then.
60 requests a second for 30 seconds: 100% request (reads).
What each component does in this design, and what calls it. The simulator mimics each one from these settings.
| Component | Its job here | Settings |
|---|---|---|
| Traffic mirror (shadow test)Traffic mirror (shadow) | Traffic enters here.Calls Load balancer, Candidate fine-tune (same GPUs).A shadow deploy copies live requests to a candidate model and compares its answers without showing them to users. | 100000 slots |
| Load balancerLoad balancer | Traffic mirror (shadow test) forwards to it.Calls AI app.Layer 4 balancers route by IP and port; layer 7 balancers read HTTP and can route by path or header. | – |
| AI appApp service | Load balancer forwards to it.Calls GPU serving pool.Little's law: throughput ≈ concurrency ÷ latency. 200 slots at 50 ms serve about 4,000 req/s. | 3 replicas × 500 slots · timeout 30 s |
| GPU serving poolSelf-hosted serving pool | AI app calls it. Candidate fine-tune (same GPUs) calls it.GPU servers batch requests together; throughput rises with batch size until memory runs out. | 72 slots |
| Candidate fine-tune (same GPUs)Multi-LoRA server | Traffic mirror (shadow test) forwards to it.Calls GPU serving pool.Multi-LoRA serving runs one base model for many fine-tuned adapters, batching requests for different adapters together. | 2000 slots |