Skip to content

The brief

Original ask (paraphrased)

Senior engineer, 15 years of experience, with about 5–6 months. I want to upgrade in software engineering, architecture, design and AI, following the latest trends. The roadmap needs resources, links, priority and complexity. Research how others build roadmaps. Include questions by complexity and use case for every topic. Structure it so I can work through it. Include newsletters and articles, and publish it as a GitHub Page I can use daily. Keep it evolving so it doesn't go stale, and update me daily. I'm good at system design and Python but want to improve everywhere. Focus on system design, Python and AI so I can implement and be productive. Don't compromise on length, but keep it structured.

Improved brief

Outcome by 14 Mar 2027 (end of week 24):

  1. Architect. Can lead the architecture of a multi-team system: C4 and ADRs, DDD boundaries, event-driven integration, migration strategy. Can defend trade-offs to executives and staff peers.
  2. AI engineer. Has shipped (in the capstone) a production-grade agentic system with:
    • a LangGraph orchestrator and Pydantic AI sub-agents
    • MCP tool servers in both Python and Spring AI
    • hybrid RAG with a reranker
    • an eval suite gating CI
    • Langfuse/OTel tracing, guardrails, and model routing with budgets
  3. Interview-ready. Scores ≥ ¾ ("Staff bar") on the unified rubric in coding, system design, AI system design and behavioral rounds at checkpoint 6.
  4. Staying current. Has a sustainable habit of about 20 min/day of reading, backed by a curated digest rather than doomscrolling.
  5. Communicates at a Staff level. Measurable gains from the week-0 baseline to week 24 in grammar accuracy, vocabulary range, spoken fluency and business writing; scores a ¾ or better on the behavioural and communication rubric rows.

Constraints:

  • Time: 12–15 h/week technical plus 30 min/day of English & communication practice (~17–18 h total). Every 6th week is lighter (weeks 6, 12, 18), plus 2 buffer weeks (25–26).
  • Budget: about $30–50/month for APIs and cloud. Default to local models (Ollama) and free tiers; use paid APIs for evals and final runs.
  • Stack:
    • Python 3.14, uv, Pydantic v2, FastAPI, LangGraph, Pydantic AI, DSPy
    • Azure (Microsoft Foundry, Container Apps, Postgres) plus Docker/Ollama locally
    • Java 25 with Spring Boot 4 and Spring AI 2.0
  • Timezone: IST. Automation runs 06:00–06:15 IST.

What the original ask was missing (now added)

Gap Added
No baseline Week 0 baseline calibrates a skip list, so you don't re-learn what you already know
No definition of done Every phase has exit gates. Every topic has a "You're done when…" line and a checklist
Reading ≠ retention Spaced repetition (DSA re-solves on days ⅓/7/21), graded L1–L4 questions with hidden answers, and Sunday explain-it-back
Interview readiness unmeasured 6 checkpoints on one rubric, with a score log to track growth
Nothing to show at the end Portfolio outputs: capstone repo, ADRs, blog posts, Staff artifacts
Burnout risk Light weeks, buffer weeks, a catch-up queue on Today (unfinished tasks from the last 7 days), and a "never drop P0, push P2" rebalancing rule
Cost blind spot Explicit budget, local-first defaults, and cost math in AI topics
"Keep it fresh" was vague Freshness contract: last_reviewed on every page, weekly version and link checks, radar changelog
Java unaddressed Parallel Java/Spring AI track that ports each capstone piece (a polyglot architecture skill)
Staff skills implicit Staff+ track with 8 concrete artifacts (strategy doc, tech-debt proposal, mentoring plan…)

Success metrics (reviewed at every checkpoint)

Metric Target by wk 12 Target by wk 24
Task completion ≥ 80% of due tasks ≥ 85%
DSA problems solved (with re-solves) 110 250
Capstone milestones M1–M5 M1–M8
ADRs written 8 15+
Mock score (avg of rounds) ≥ 2.5 / 4 ≥ 3.0 / 4
Longest streak 21 days 45 days