Week 14 — Cost, latency, routing¶

At a glance

Phase 4: Production & AI Architecture · Starts: Mon 28 Dec 2026 · Planned: 16h 45m

Build: Capstone: LiteLLM routing (cheap → strong escalation), semantic + prompt caching, streaming latency budget

By day¶

System Design · 45 min · SLOs, SLIs, error budgets, burn-rate alerting — define SLOs for the capstone (latency, success, eval-quality SLO) → Observability, SLOs & error budgets w14-sd-1
DSA · 25 min · Redundant Connection (union-find) → Graphs: BFS, DFS, topological sort, union-find · resource w14-dsa-1
Python · 20 min · Rep: OTel metrics (histograms for token usage, latency) exported to Langfuse/Prometheus → Observability in Python: structlog, OpenTelemetry w14-py-1
Communication · 30 min · Grammar: Degrees of certainty: modals and adverbs → Week 14 drills · day 1 w14-comm-1

Agentic AI · 60 min · Cost & latency levers: prompt caching, semantic caching, batching, streaming, speculative/parallel calls — build a latency budget per node → Cost & latency optimization: caching, batching, streaming · resource w14-ai-1
DSA · 30 min · Word Ladder → Graphs: BFS, DFS, topological sort, union-find · resource w14-dsa-2
Communication · 30 min · Vocabulary: Negotiation vocabulary: concession, leverage, deadlock → Week 14 drills · day 2 w14-comm-2

DSA · 25 min · Network Delay Time (Dijkstra) → Advanced graphs: Dijkstra, MST, Bellman-Ford · resource w14-dsa-3
Architecture · 45 min · C4 model + arc42: draw capstone context, container, component diagrams as code (Structurizr DSL or Mermaid) → C4, arc42 & diagrams as code · resource w14-arch-1
Python · 20 min · Rep: async token-bucket rate limiter for provider calls → asyncio in depth: TaskGroups, cancellation, backpressure w14-py-2
Communication · 30 min · Speaking drill: Handling pushback live → Week 14 drills · day 3 w14-comm-3

Agentic AI · 60 min · Model selection & routing: cascades, classifier routers, fallbacks; LiteLLM router config with budgets → Model selection, routing & gateways (LiteLLM) · resource w14-ai-2
Java/Spring AI · 30 min · Export Spring AI OTel traces to Langfuse; align attributes with GenAI semconv → Observability & testing: Micrometer, OTel, Testcontainers · resource w14-java-1
Communication · 30 min · Idioms & phrasal verbs: Negotiation phrasal verbs: hold out, give in, back down → Week 14 drills · day 4 w14-comm-4

System Design · 45 min · LLM capacity & cost planning: tokens/s, TTFT, concurrency, $ per resolved incident — spreadsheet model → LLM capacity, latency & cost planning w14-sd-2
DSA · 25 min · Min Cost to Connect All Points (Prim/Kruskal) → Advanced graphs: Dijkstra, MST, Bellman-Ford · resource w14-dsa-4
Python · 20 min · Rep: snapshot tests for prompts + cost assertions in pytest → Testing: pytest, fixtures, Hypothesis, testcontainers w14-py-3
Communication · 30 min · Writing: Writing a proposal → Week 14 drills · day 5 w14-comm-5

Agentic AI · 120 min · Capstone build: LiteLLM proxy in docker-compose (Ollama + 2 hosted models), cascade routing with eval-guarded escalation, prompt caching; report cost/latency before vs after → Model selection, routing & gateways (LiteLLM) · resource w14-ai-3
DSA · 30 min · Cheapest Flights Within K Stops (Bellman-Ford / BFS with stops) → Advanced graphs: Dijkstra, MST, Bellman-Ford · resource w14-dsa-5
Java/Spring AI · 30 min · Spring AI guardrail advisor (input/output check) in the Java service → Agentic patterns with Spring AI w14-java-2
Communication · 30 min · Record & shadow: Handling pushback live → Week 14 drills · day 6 w14-comm-6

System Design · 45 min · Written design: payment system — idempotency, ledger (double-entry), exactly-once illusions, reconciliation, PSP integration → Payment system w14-sd-3
Architecture · 30 min · Write ADR 014 for capstone: LiteLLM as model gateway (vs custom router vs managed gateway) → AI-native architecture: LLMs as system components · resource w14-arch-2
Staff+ · 30 min · Artifact: technical strategy doc (2-3 pages) — 'AI agents for operations: diagnosis, guiding policies, coherent actions' → Writing engineering strategy & vision · resource w14-staff-1
Communication · 30 min · Soft skills + weekly review: Negotiation practice → Week 14 drills · day 7 w14-comm-7
Review · 60 min · Run a mock architecture review of the capstone C4 docs + ADRs with a peer/AI reviewer: checklist, risks, decision log; flashcards on SLOs and graph algorithms → Running architecture reviews & guilds w14-rev-1