The job market shifted — your roadmap should too

The “learn to code in three months and earn six figures” era is over. By April 2026, AI coding assistants have squeezed the entry-level market hard, particularly for purely-CRUD work that bootcamp graduates traditionally targeted. The good news: the bar moved, but the door isn’t closed. Companies still hire entry-level engineers — they just expect more system understanding, more communication, and more demonstrated ownership than the 2022 era required.

This roadmap is aimed at someone starting from zero with 12–15 hours per week to spare and a 12-month horizon. It’s calibrated to what’s actually being asked in interviews this quarter.

The 12-month plan at a glance

PhaseMonthsFocusHoursOutput
Foundations1–2Computer fundamentals, one language605–7 small programs
Full-stack base3–5Web dev, databases, Git, deployment901 portfolio project
Systems and depth6–8Data structures, async, testing901 substantial project
Specialization9–10Pick a domain (web, data, mobile, ML)601 specialist project
Job hunt11–12Resume, interviews, applications6030+ applications

Total: about 360 hours across 12 months. That’s 7 hours per week, sustainable around a day job.

Months 1–2: foundations that don’t get skipped

Pick one language and stick with it. In 2026 the best entry-level languages are still Python (data, ML, scripting), JavaScript/TypeScript (web), and Go (backend). For absolute beginners, Python is the most forgiving.

You’re not learning the language — you’re learning how to think about computers. Topics that matter:

  • Variables, control flow, functions, scope (you’ll re-learn these in every language anyway)
  • The terminal — cd, ls, grep, basic shell scripting
  • Git — at least clone, add, commit, push, pull, branch, merge
  • How HTTP works at a 10,000-foot level
  • One short OS book — Operating Systems: Three Easy Pieces (free online)

Five small programs: a unit converter, a quiz game, a CSV parser, a weather CLI, a habit tracker.

Months 3–5: full-stack base

Now build something a stranger can use. The minimum technical surface for a 2026 entry-level web role:

  • HTML/CSS, modern responsive layout (Flexbox, Grid)
  • A frontend framework (React or Svelte — pick one, learn deeply)
  • A backend (Node.js + Express, or Python + FastAPI, or Go + Echo)
  • A relational database (Postgres) and one query language (SQL)
  • Deployment to a real URL (Cloudflare Pages, Vercel, Fly.io, Render)
  • HTTP fundamentals (status codes, REST, JSON, auth basics)

Build one full-stack portfolio project: an app you’d actually use. A budget tracker, a workout log, a habit streak app. Deploy it. Show it works.

Months 6–8: systems and depth

This is where bootcamp graduates often plateau and self-taught engineers can pull ahead. Topics:

  • Data structures and algorithms (LeetCode Easy → Medium, ~50 problems)
  • How databases really work (indexes, transactions, joins, query plans)
  • Async/concurrency (event loop in JS, asyncio in Python, goroutines in Go)
  • Testing (unit, integration; one framework like Pytest or Vitest)
  • Caching, queues, and basic system design vocabulary

Build a substantial project that demonstrates depth. Real-time chat, a scraper with scheduled jobs, a CSV-to-API SaaS prototype. Use real services (Postgres, Redis, an SQS-style queue), not just localhost.

Months 9–10: specialization

Generalists struggle to stand out. Pick a lane and lean in:

  • Web (frontend) — React deep dive, accessibility, design systems
  • Web (backend) — system design, scaling patterns, observability
  • Data engineering — pipelines, dbt, basic Spark, SQL deep dive
  • Mobile — React Native or Swift / Kotlin
  • ML engineering — PyTorch, fine-tuning, evals, basic MLOps
  • DevOps/Platform — Docker, Kubernetes, CI/CD, infrastructure as code

Build one specialist project that demonstrates this lane. A real-time dashboard, a working CDC pipeline, a published mobile app, a fine-tuned local model.

Months 11–12: the job hunt

The worst self-taught engineers wait until month 12 to start applying. The best start in month 9. Why: cold applications take 4–8 weeks to convert, so applying late means a multi-month income gap.

A realistic job-hunt setup for 2026:

  1. Resume on a single page — your projects, links to live demos, GitHub username
  2. GitHub profile with 4–5 strong repos, README explaining each
  3. Personal site with one paragraph, links to projects, contact info
  4. 30+ applications minimum — personalized cover letters cut response rate, but boilerplate scales
  5. 2–3 referrals from your network — convert a 4% reply rate into a 30%+ rate

What employers actually ask in 2026 entry-level interviews

Conversations with hiring managers across SaaS, fintech, and devtools companies in Q1 2026 surface a consistent pattern:

  • One coding problem (Easy / Medium LeetCode level)
  • One system design walkthrough at a beginner depth (“how would you build a URL shortener?”)
  • One project deep-dive (“walk me through your hardest technical decision in this repo”)
  • One AI-tools competency check (“how do you use Cursor / Copilot / Claude Code in your workflow?”)

The last one is new. In 2026, fluency with AI coding tools is expected. Companies are not impressed by candidates who avoid them, nor by candidates who can’t critique their output. Demonstrate that you use them and review their code.

The five most common self-taught mistakes

  1. Tutorial hell — watching videos endlessly without building. Build first, debug, then watch tutorials when stuck.
  2. Too many languages — pick one for the first 6 months. Polyglot LinkedIn looks impressive and signals nothing.
  3. No deployed projects — a project that’s not live is worth half a project. Push it to a real URL.
  4. Skipping fundamentals — you cannot fake it past the systems-and-depth phase. Invest the time.
  5. Avoiding interviews until you’re “ready” — you’re never ready. Start interviewing two months before you think you should.

Realistic outcome — what to expect

12 months of consistent self-study, with the right project portfolio and network, gets most people to:

  • 5–15 first-round interviews
  • 1–3 final-round interviews
  • 1 offer in the $55,000–$95,000 USD range (US, remote-friendly)

Outside the US the range varies widely — research your local market via levels.fyi and Stack Overflow Developer Survey. Be patient: the second job typically pays 30–50% more after 12–18 months of professional experience.

Disclosure

This article contains affiliate links to learning platforms. We may earn a commission at no extra cost to you. Salary ranges are general estimates based on public data and vary by region, role, and company.

Sources