Strong basics
DSA, OOP, system design.
Is AI going to replace software engineers?
Coding ≠ Engineering
Yes, if you stay a coder.
No, if you become an engineer.
Every one of us starts as a coder. I did too. The real question: will you stop there?
Who's a coder? Who's an engineer?
Player 1
Coder
Gets a ticket. Writes the code. Done.
AI does this part very well now.
VS
Player 2
Engineer
Owns the problem, not just the code.
+1000 XP
What makes you an engineer?
Ownership. Of what, bro?
- 🏆Picking up a business problem
- 🧩Breaking it down into smaller problems
- 🔍Finding the root cause
- 🏗️Designing and building the solution
How do you build this? Put yourself in front of real problems, again and again. Fall in love with solving problems, not with writing code.
The role that's booming
−0%
Fresher hiring at big tech since 2019
+0%
Forward Deployed Engineer postings in one year
Salesforce alone wants 1,000 FDEs. Palantir, Databricks and EY are building FDE teams.
FDE
PM
Dev
QA
Presenter
Goes to the client's office, picks up their problem and solves it end to end. Companies didn't stop hiring engineers. They stopped hiring people who only take orders.
Change is the only constant
We're in the AI era. To be exact, the AI bubble era.
Plus $518B planned on cloud and data centres. To be fair, they made their first operating profit in Q2 2026 ($11.5B revenue). The business is real. The valuation is the bubble.
It burst while the internet was growing. The internet still became everyone's normal.
Even if the AI bubble bursts, AI will be the new normal.
Did AI kill tech jobs?
- 280,316 open tech jobs, Sep 30 2026
- +72% from the early-2023 low
- −41% from the 2022 peak
- Claude Code came. Opus 4.5 came. Openings went up.
- The 2022 crash was companies fixing their over-hiring. Not AI.
Source: trueup.io
Where AI actually helps
From my own work. My numbers, not industry data.
- Planning2x
- Writing code10–20x
- Testing1x
- Deployment3x
Now the twist. Researchers at METR measured it properly.
2025: experienced devs with AI were 19% slower. They felt 24% faster.
Early 2026: about 18% faster.
AI multiplies what you already know. Zero into anything is still zero.
Anthropic makes Claude. Their CEO says engineers there "don't write code anymore". Yet they have 1,000+ open roles, and engineering is their biggest team. Writing code got automated. Judgement didn't. AI is not taking over us. It's helping us.
Live demo
claude
> █
🐛 Two people can book the same seat.
🐛 No index on the table that gets searched the most.
🐛 No input validation.
Learn Claude Code. Learn Codex. But use AI to write code only if you can write it yourself. Otherwise you won't catch what I just caught.
What companies actually look for
Ownership
Problem, solution, shipped. The FDE mindset.
AI with judgement
Go fast with AI. Then review, debug and verify.
Communication
Explain why, not just what.
Proof, not certificates
Projects that are live, with a link.
What should I learn?
System design.
System design.
System design.
- LLD, hands-on
- Architecture
- OOPs
- DSA is never dead. Interviews still ask it, and you need it to spot when AI hands you an O(n²) mess.
- RAG, agentic AI, agent frameworks. Don't just read about them. Build one small agent.
Solving a system design problem
The steps, every single time.
- Functional requirementsWhat must it do?
- Non-functional requirementsHow fast, how big, how available? CAP lives here.
- Tech stackPick tools for the problem, not for your CV.
- Back-of-the-envelopeUsers, requests per second, storage.
- Low-level architectureBoxes, arrows, who talks to whom.
- API designThe contract.
- DB designTables, keys, indexes.
Concepts to explore 📸 take a screenshot
- Rate limiting
- Sharding
- Caching
- DB replication
- Async processing
- Queues
- Fan-out pattern
- Fan-in pattern
- Bottlenecks
- Batch jobs
- CDN
- Outbox pattern
- CAP theorem
- Eventual consistency
- Storage tiering
- SAGA
- Transactions
Warning: boss level
30-day challenge
30days
- Build one real project and deploy it. End to end. Use Claude Code or Codex.
- Write down every decision and every bug. And why.
- Put it on GitHub with a proper README and a live link.
- Explain its system design out loud. To a friend, or to your phone camera.
In 30 days you'll have something most of your batch won't: proof.
Mindset matters more than you think
Go after Plan A like there is no Plan B.
If you don't try, you've already lost.
- What you think, and what you do daily, is what you become.
- Jack of all trades, master of one. Know a little of everything. Go deep in one thing.
- Even water goes bad when it stops moving. Only running water stays useful.
- Someone who fears failure more than pain will never learn. Taste both.
Try. Fail. Unlearn. Relearn. Learn. Grow.
?
Questions?
Ask anything. Nothing is a silly question here.
All the very best
For whatever you build next. Go make something real.