Week 5: Heap / Priority Queue · Backtracking

Target times: Medium ≤ 25m · Hard ≤ 45m  ·  Day 1 = Monday

~2 hrs/workday: 75 min DSA (timed, no AI autocomplete) + 10 min syntax typing + 35 min second block. Weekend 3–4 hrs: project + long DSA + 1 system-design practice out loud. Over cap → 48h redo queue.

Day 1 · Mon

DSA set 1 of 5

DSA problems (75 min, timed)

Second block (35 min) — System Design + CS — System Design Core I:
Load balancing (L4/L7), horizontal scaling

Day 2 · Tue

DSA set 2 of 5

DSA problems (75 min, timed)

Second block (35 min) — ML / DL / LLM / VLM — CNNs (your CV domain):
Convolution, pooling, receptive field

Day 3 · Wed

DSA set 3 of 5

DSA problems (75 min, timed)

Second block (35 min) — System Design + CS — System Design Core I:
Caching strategies, eviction, CDN

Day 4 · Thu

DSA set 4 of 5

DSA problems (75 min, timed)

Second block (35 min) — ML / DL / LLM / VLM — CNNs (your CV domain):
Architectures: ResNet, EfficientNet/Det

Day 5 · Fri

DSA set 5 of 5

DSA problems (75 min, timed)

Second block (35 min) — Second block — mix:
Back-of-envelope estimation (QPS, storage)  •  Transfer learning, augmentation

Weekend · Sat–Sun

3–4 hrs/day
🚀 Project 2 — Multi-Model Inference Server + Benchmark — KICKOFF: Serve 2–3 models (FastAPI multi-model). Build a benchmark harness (Locust/custom) for latency + throughput under concurrency. Milestone: baseline server + benchmark running.