Week 1: Python Reset + DSA Launch

Note: Drill from memory, no autocomplete. 10 min SyntaxCache/SpeedCoder. Outcome: write list ops + comprehensions without lookup.

Note: deque = your BFS/queue workhorse. Outcome: write defaultdict/Counter/deque from memory.

Note: Watch NeetCode 'Python for Coding Interviews'. Outcome: heapq + string ops from memory; accounts ready.

Second block (35 min) — System Design + CS: OS Basics
Process vs thread, context switching, concurrency vs parallelism, virtual memory, deadlock (4 conditions). Most important: process vs thread + why concurrency matters. Read OSTEP intro / Neso Academy OS (first 5). Outcome: explain process vs thread + a deadlock scenario.

Second block (35 min) — ML/DL: ML Metrics & bias-variance
Precision/recall/F1, ROC-AUC vs PR-AUC, bias-variance, overfitting, regularization (L1/L2). Most important: PR-AUC vs ROC-AUC on imbalanced data (your medical-imaging domain). Watch StatQuest. Outcome: explain when PR-AUC beats ROC-AUC.

Second block (35 min) — System Design + CS: OS Basics
Process vs thread, context switching, concurrency vs parallelism, virtual memory, deadlock (4 conditions). Most important: process vs thread + why concurrency matters. Read OSTEP intro / Neso Academy OS (first 5). Outcome: explain process vs thread + a deadlock scenario.

Second block (35 min) — ML/DL: ML Metrics & bias-variance
Precision/recall/F1, ROC-AUC vs PR-AUC, bias-variance, overfitting, regularization (L1/L2). Most important: PR-AUC vs ROC-AUC on imbalanced data (your medical-imaging domain). Watch StatQuest. Outcome: explain when PR-AUC beats ROC-AUC.