Frontend component building (Next.js/Tailwind)

AI code generators like v0, Cursor, and Copilot already scaffold entire React/Next.js UIs from a prompt, making junior frontend work largely commoditized.

Basic fullstack CRUD development

Standard CRUD apps, API wiring, and boilerplate integrations are the exact tasks LLM coding agents (Devin, Claude Code) are optimized to replace first.

Integrating third-party libraries/tools

Wiring together FFmpeg, Tesseract OCR, and LibreOffice is exactly the kind of glue-code task AI agents can generate and debug autonomously with enough context.

Junior-level debugging

AI assistants can now trace stack errors and suggest fixes faster than a 2-year dev without deep systems context.

Computer Vision & NLP coursework

If pursued seriously beyond coursework, deep ML/CV expertise is harder to automate than typical web dev work and could pivot him toward building AI tools instead of being replaced by them.

Open-source project ownership

Shipping and maintaining a full product (not just tickets) shows end-to-end judgment, architecture decisions, and product sense that AI still struggles to replicate independently.

Multilingual communication

Fluency in five languages is useful for client-facing or localization-heavy roles that require nuanced human context AI often mishandles.