Core hours and optional add-ons
The starting estimate should use the core path only. Round mock bootcamps, Round 2 depth, and camp prep are separate add-ons based on the student's target and timeline.
| Module | Core topics | Est. hours |
|---|---|---|
| Core shared base | ||
| F1 Foundations I | NumPy/Pandas/scikit-learn — regression, classification, clustering | ~10 hrs |
| F2 Foundations II | PyTorch — MLP and a mini transformer, built from scratch | ~14 hrs |
| A1 Computer Vision | CNN to diffusion — classify, detect, segment, generate | ~10 hrs |
| A2 NLP & Audio | BERT, LLM APIs, Whisper — text and audio end-to-end | ~10 hrs |
| USAAIO format core | ||
| U0 Eligibility + registration | Citizenship/visa check, portal/account/proctor-site setup | ~2 hrs |
| U1 Math + Colab boost | By-hand math derivations + standardized Colab markdown answers | ~10 hrs |
| Optional add-ons after target is clear | ||
| U2 Round 1 Mock Bootcamp | 3-hour / CPU-only format — stamina, speed, accuracy, error log | +~10 hrs |
| U3 Round 2 Bootcamp | MIT / GPU L4 env — Transformers, NLP, CV, generative practice | +~13 hrs |
| U4 Camp prep | Team USA selection simulation + IOAI/IAIO finals exposure | +~10 hrs |
Timeline by starting point
Lower weekly load and more time for public project evidence.
Build the shared base and learn notebook answer discipline.
Compressed preparation; target choice matters earlier.
Early-start advantage
※ Hours are 1:1 recommendations and vary ±30–50% per student. Optional add-ons are not fixed requirements. Dual-track students reuse the F/A base instead of repeating it; the exact starting module is confirmed after the level test.