Video summary
Behind the Code - PodCast with Belal Khan @SimplifiedCoding Freshers| SDE1s | JOB | AI | ANDROID
Main summary
Key takeaways
Key technological concepts, product/tooling views, and practical guidance
AI vs “AI engineering” (portfolio expectations)
- Bilal argues that writing/using AI-generated code is not the same as “AI engineering.”
- Companies may encourage developers to generate the first code version with AI, but this can lead to “garbage code” accumulation unless reviews are done carefully.
- For freshers/students, he recommends avoiding AI for learning fundamentals first—without basic understanding, you can’t verify correctness or craft good prompts.
- He also counters a market myth: using ChatGPT doesn’t automatically make you an AI engineer.
Learning Android now: how to use AI while still building skill
- He suggests using AI mainly to speed up building, while still learning:
- architecture-level decisions
- fundamentals for correct solutions
- how to ask better questions and articulate UX requirements
- In prompt/workflow quality, articulation matters as much as coding (poor prompts → poor outputs).
- He notes that Compose UI generation and design-language alignment may require extra work beyond generic prompting.
Android tooling transition (Eclipse → Android Studio → AI-assisted tools)
- For learning Android, he strongly advises starting with Android Studio.
- He mentions AI coding environments (e.g., “Antigravity Cloud” / agent mode in his company context):
- In companies, AI tools may be restricted; only certain agents/models might be available.
- For learning, AI-heavy tools should generally be optional, not primary.
Personal projects: AI speeds up early screens, but you still need engineering decisions
- AI can quickly generate multiple screens, but projects often stall afterward unless you know what to ask next.
- The implication: AI reduces boilerplate, but product thinking and system design remain the differentiator.
Android ecosystem changes (developer friction vs user benefit)
- “Android is becoming like iOS” in app publishing complexity:
- longer Play Store review
- more policy compliance (e.g., privacy policy, account deletion)
- Mentions permission/access restrictions for developer capabilities (e.g., earlier SMS-related access).
- UI/compatibility requirements are increasing:
- movement toward adaptive layouts (e.g., avoiding hard locks to portrait-only)
- more developer effort to support device constraints
Device fragmentation & production bugs
- Challenges include:
- OEM/custom OS behavior affecting background scheduling
- many screen sizes/resolutions, requiring responsive layouts
- production bugs that may be invisible on some devices but show up through user reports
UI development: XML → Jetpack Compose
- He’s enthusiastic about Compose:
- new work increasingly becomes “composed”
- faster UI creation
- easier alignment when using Figma-to-UI workflows
- He notes legacy modules still exist.
- Compose isn’t truly “no-code” when you need reusable patterns or a custom design language.
Job market perspective (Android roles & hiring trends)
- Large companies often hire for clear fundamentals, not only “specific Android dev.”
- On-campus hiring tends to focus on learning ability + fundamentals.
- Off-campus paths require strong projects and proof.
- Despite “market is bad” narratives, he claims jobs still happen and recruiters reach out (based on his experience).
What should be in a fresher/tier-3 portfolio now (post-AI era)
- Old advice (“create repos/projects”) is less effective because everyone can generate apps/projects with AI.
- Differentiators now:
- architecture-level understanding
- clear tradeoffs and decisions
- strong fundamentals (e.g., DSA/LeetCode remains important—especially for senior roles and remote interviews)
- “Showing an app” isn’t enough because many can build similar apps quickly.
Engineering workflow maturity: PR reviews as learning
- He recommends learning via end-to-end projects, not toy/shelved apps.
- He emphasizes PR-based workflows:
- seniors review code
- discussions teach architecture, corrections, and better feature planning
- Startups may move faster with less PR review, which reduces this learning curve.
Kotlin Multiplatform (KMP) view
- He isn’t using KMP in his current workplace, but previously freelanced a product with it.
- Past issues included:
- iOS compilation/UI rendering hiccups
- slower builds
- He believes KMP is improving and that more companies adopt it for business reasons:
- shared business logic in a common codebase
- reduced team size via one cross-platform effort
- He expects more maturity over the next few years.
Google Developer Expert (GDx) and community value
- He became Google Developer Expert around 2019 (with videos dating back to 2014).
- He credits GDx with helping him:
- build a network
- stay aligned with what Google is working on
- gain opportunities via referrals/contacts
Practical “rules” he repeatedly implies for beginners/freshers
- Learn foundations first; use AI later (or carefully) once fundamentals are solid.
- Don’t just ship apps—demonstrate architecture decisions, tradeoffs, and system reasoning.
- Expect DSA/LeetCode and coding tests to remain relevant.
- Prefer learning through end-to-end project completion and code review/PR discussions.
Main speakers / sources
- Bilal Khan — Android Developer at American Express, creator of Simplified Coding, Google Developer Expert (GDx).
- Rohit — interviewer/host; speaks throughout the podcast.