Week 11 — Durable execution, HITL, multi-agent¶
At a glance
Phase 3: Agents + Protocols · Starts: Mon 07 Dec 2026 · Planned: 16h 45m
Build: Capstone: human-in-the-loop approval (interrupt/resume) + Spring AI MCP server wired in
By day¶
Monday 07 Dec · 2h 00m¶
- System Design · 45 min · Circuit breakers, bulkheads, load shedding, timeouts hierarchy — apply to tool calls and LLM providers → Reliability: retries, backoff, circuit breakers, bulkheads
w11-sd-1 - DSA · 25 min · Kth Largest Element in an Array (quickselect vs heap) → Heap / priority queue · resource
w11-dsa-1 - Python · 20 min · Rep: uv workspaces monorepo — orchestrator, mcp-server, evals as members → Packaging, project layout & monorepos · resource
w11-py-1 - Communication · 30 min · Grammar: Advanced articles and the zero article with abstract nouns → Week 11 drills · day 1
w11-comm-1
Tuesday 08 Dec · 2h 00m¶
- Agentic AI · 60 min · Durable execution & HITL: LangGraph interrupt/Command resume, idempotent side effects, timeouts; compare with Temporal-style durable workflows → Durable execution & human-in-the-loop · resource
w11-ai-1 - DSA · 30 min · Task Scheduler → Heap / priority queue · resource
w11-dsa-2 - Communication · 30 min · Vocabulary: Register: formal vs informal, and switching → Week 11 drills · day 2
w11-comm-2
Wednesday 09 Dec · 2h 00m¶
- DSA · 25 min · Design Twitter + Find Median from Data Stream → Heap / priority queue · resource
w11-dsa-3 - Architecture · 45 min · Enterprise integration patterns: channels, content-based router, aggregator, idempotent receiver — relate to MCP/A2A → Enterprise integration patterns
w11-arch-1 - Python · 20 min · Rep: anyio/pytest-asyncio tests for the interrupt/resume flow → Testing: pytest, fixtures, Hypothesis, testcontainers
w11-py-2 - Communication · 30 min · Speaking drill: Clarity on calls and video: pacing, signposting → Week 11 drills · day 3
w11-comm-3
Thursday 10 Dec · 2h 00m¶
- Agentic AI · 60 min · Multi-agent systems: supervisor vs swarm vs hierarchical; failure modes (loops, context loss, cost blow-up) — decide if capstone needs >1 agent → Multi-agent systems: when, how, and failure modes · resource
w11-ai-2 - Java/Spring AI · 30 min · Finish Spring AI MCP server: OAuth-ready config, tool schemas, integration test from the Python LangGraph client → Tool calling & MCP servers in Spring AI
w11-java-1 - Communication · 30 min · Idioms & phrasal verbs: Escalation phrasal verbs: flag up, push back, loop in → Week 11 drills · day 4
w11-comm-4
Friday 11 Dec · 2h 00m¶
- System Design · 45 min · Distributed job scheduler: leases, at-least-once execution, cron fan-out, dedupe — sketch + read notes → Distributed job scheduler
w11-sd-2 - DSA · 25 min · Subsets → Backtracking · resource
w11-dsa-4 - Python · 20 min · Rep: DuckDB over Langfuse trace exports to find slowest nodes → Data tooling: Polars, DuckDB, Arrow
w11-py-3 - Communication · 30 min · Writing: Escalation emails → Week 11 drills · day 5
w11-comm-5
Saturday 12 Dec · 3h 30m¶
- Agentic AI · 120 min · Capstone build: HITL approval node before create_ticket (interrupt → approve via API → resume from checkpoint) calling the Spring AI MCP server; kill/restart mid-run to prove durability → Durable execution & human-in-the-loop
w11-ai-3 - DSA · 30 min · Combination Sum → Backtracking · resource
w11-dsa-5 - Java/Spring AI · 30 min · Spring AI agentic patterns: chain, routing, orchestrator-workers examples — port one to the Java service → Agentic patterns with Spring AI · resource
w11-java-2 - Communication · 30 min · Record & shadow: Clarity on calls and video: pacing, signposting → Week 11 drills · day 6
w11-comm-6
Sunday 13 Dec · 3h 15m¶
- System Design · 45 min · Written design: multi-agent platform — agent registry, tool gateway (MCP), memory, orchestration, evals, tenancy, cost controls → Design a multi-agent platform
w11-sd-3 - Architecture · 30 min · Write ADR 011 for capstone: MCP as integration boundary for tools (vs direct SDK clients) → Enterprise integration patterns · resource
w11-arch-2 - Staff+ · 30 min · Draft an exec update for your manager's manager on the capstone/AI initiative: 5 bullets, 1 chart, 1 ask → Communicating with executives & stakeholders
w11-staff-1 - Communication · 30 min · Soft skills + weekly review: Saying no without burning bridges → Week 11 drills · day 7
w11-comm-7 - Review · 60 min · Retro + explain-it-back: interrupt/resume guarantees and when multi-agent is worth it; flashcards on circuit breakers → Multi-agent systems: when, how, and failure modes
w11-rev-1
By track¶
| Track | Tasks | Time |
|---|---|---|
| Agentic AI | 3 | 4h 00m |
| System Design | 3 | 2h 15m |
| DSA | 5 | 2h 15m |
| Architecture | 2 | 1h 15m |
| Python | 3 | 1h 00m |
| Java/Spring AI | 2 | 1h 00m |
| Staff+ | 1 | 30m |
| Communication | 7 | 3h 30m |
| Review | 1 | 1h 00m |
End-of-week¶
- Weekly retro written in
docs/log/retros/(template: retro) - Flashcards / explain-it-back done for this week's topics
- Progress synced (close the daily GitHub issues)