"Writing Code Is Getting Commoditised, Not Engineering" — Arsh Goyal at DevSparks Pune

"Writing Code Is Getting Commoditised, Not Engineering" — Arsh Goyal at DevSparks Pune

Every software developer in India is asking some version of the same question right now: Will AI take my job? At DevSparks Pune 2026 — India's flagship developer summit held on February 28, 2026 — AI influencer and engineer Arsh Goyal gave the most direct, thought-provoking answer you'll hear: sort of, but not in the way you think.

His talk was titled "From Writing to Directing Code: The New Software Engineer" — and the core argument deserves unpacking for every developer, student, and IT professional in India.

The Core Argument: Code ≠ Engineering#

Goyal's thesis is sharp and practical. He separates two things that most people conflate:

• Writing code — translating a known requirement into working syntax. This is increasingly something AI can do.
• Engineering — understanding systems, making architectural decisions, navigating ambiguity, thinking about trade-offs, managing complexity. This remains distinctly human.

The analogy that resonates: a word processor commoditised typing. But it didn't commoditise writing — the craft, the ideas, the structure. Similarly, GitHub Copilot, Cursor, and Claude can generate thousands of lines of syntactically correct code. But they can't decide whether a microservices architecture makes sense for your startup's scale. They can't evaluate whether that database schema will survive 10x user growth. They don't understand your company's technical debt history.

What DevSparks Pune Was About#

DevSparks Pune 2026 brought together developers, Global Capability Center (GCC) leaders, product thinkers, and AI practitioners for a day of hard conversations about where software engineering is heading. Topics included coding agents, agentic AI, multi-agent orchestration, and generative AI for productivity. The audience was largely professional developers and engineering leads — people who've already started using AI tools and are now asking what comes next.

Goyal's talk arrived at exactly the right moment. With tools like Cursor, GitHub Copilot, and Devin (AI software engineer) becoming mainstream in 2025, the developer anxiety is real and justified — not paranoid.

The "Directing Code" Mental Model#

The reframe Goyal offers is valuable: stop thinking of yourself as a writer of code and start thinking of yourself as a director of code. A film director doesn't operate the camera — they decide what story to tell, how to tell it, what shots serve the narrative, and they guide a team of specialists toward a coherent vision.

An engineer in the AI era:

• Defines the problem clearly enough that AI can solve the right thing, not just any thing
• Reviews AI-generated code for security vulnerabilities, performance issues, and architectural misalignment
• Understands the system holistically — AI tools see locally, engineers think globally
• Makes judgment calls on ambiguous trade-offs that have no objectively correct answer
• Communicates technical complexity to non-technical stakeholders

What Indian Developers Need to Do Right Now#

India has 5+ million software developers — the world's second-largest developer pool. The implications of Goyal's argument are direct:

1. Stop competing on typing speed, start competing on thinking quality#

If your value as a developer is "I write code fast," you're in the most vulnerable position. The developers who thrive will be the ones who can define requirements precisely, spot bad architecture before it becomes technical debt, and understand systems at a level AI tools can't.

2. Learn to use AI tools fluently — as a multiplier, not a crutch#

Developers who use Copilot, Cursor, and Claude effectively are already 3-5x more productive than those who don't. But the ones who just copy-paste AI output without understanding it are accumulating hidden technical debt. Use AI to move fast, but understand what it generates.

3. Invest in system design and architecture skills#

The interview question that was always hard — "Design a system like YouTube at scale" — is now the most important skill differentiation. System design, distributed systems, database architecture, security architecture — these are the things AI tools genuinely struggle with at the holistic level.

4. Communication and stakeholder management become technical skills#

If AI handles more of the actual coding, engineers who can translate business problems into technical requirements (and explain technical constraints back to business) become enormously valuable. This has always been true at senior levels — it's now true earlier in careers.

The Jobs Picture for India#

The honest answer is: some coding jobs will disappear, new ones will be created, and the remaining ones will require different skills. Historical analogies help — the ATM didn't eliminate bank teller jobs, it changed what tellers do (higher-value customer interaction). But the transition was painful for those who didn't adapt.

India's IT outsourcing model — where value was partly in labor arbitrage for routine coding work — faces the most disruption. But India's strength in engineering talent, GCC growth, and product-company emergence creates real opportunities for those who move up the value chain.

Arsh Goyal's Takeaway#

Goyal's closing message at DevSparks was essentially this: don't be afraid of AI taking your coding job. Be afraid of another developer using AI to do your job better than you. The threat isn't AI replacing engineers — it's AI-augmented engineers replacing non-AI-augmented engineers. The tool is available to everyone. The question is who wields it most skillfully.