I had the vision. I had the architecture. I just didn’t have the hands and time — until now.
The Lie We Tell Ourselves
As a lead — did you ever come across a situation where you spent your entire day playing around delegation and consolidation rather than really writing code, building prototypes, and solving the problems you were actually hired to solve?
Be honest.
Sprint planning. Backlog grooming. Stakeholder alignment. One-on-ones. Escalation calls. Architecture reviews. The calendar is a wall of colour-coded blocks with no white space.
By 6 PM, you’ve managed everything — and built nothing.
And then the real work begins.
Because the bugs don’t fix themselves at 6 PM. The code doesn’t write itself while you were explaining the same architecture to the third stakeholder today. The feature your team couldn’t finish? It’s still sitting there. Waiting for the one person who holds the full picture.
You.
So you open your laptop at 9 PM. Pour another coffee. And start doing the work that delegation was supposed to handle — but never quite did.
I did this for fifteen years.
Night after night. Sprint after sprint. I told myself it was leadership. It was ownership. It was “the price you pay for being senior.”
But here’s the truth nobody talks about: even when you do delegation right, it’s still broken.
You write detailed Jira tickets. You add acceptance criteria. You attach architecture diagrams. You explain the context in a refinement session. You answer follow-up questions on Slack. You do a walkthrough call. You pair program the first module together.
And after all of that — you still don’t get 100% of what you envisioned.
Not because the team is bad. They’re talented. But the developer who picks up the ticket might not have the depth in that particular security pattern. Or the timeline is too tight to explore the edge cases you would have explored. Or they interpreted the requirement slightly differently — and by the time you catch it in code review, half the sprint is gone and rework isn’t an option.
The gap between what a lead sees and what a team delivers is not a people problem. It’s a physics problem. The full picture — functional, non-functional, security, scalability, coding standards, deployment nuances — lives in one brain. And no amount of Jira formatting, story pointing, or documentation can transfer it without loss.
So I became the safety net. The architect by day. The developer by night. The lead who led meetings until 6 PM and debugged until midnight.
I wore it like a badge of honour.
It was a chain.
And the root cause? Delegation is busted badly. Not because people can’t follow instructions — but because information in transition tends to lose some of its part. Either while updating the Jira ticket or while interpreting the task. The full picture in a lead’s head can never materialise fully, no matter how detailed you make it.
The Dream That Wouldn’t Die
Here’s the thing about being a lead who codes at midnight — it doesn’t just steal your sleep. It steals your ambition.
Because somewhere behind the sprint boards and escalation calls and 11 PM debugging sessions, there was a dream. A side venture. A hustle. Something I wanted to build — not for a corporation, not for a client, not for a Jira ticket — but for myself.
VedicJivan. A digital platform for Vedic astrology and spiritual wellness. I had the complete vision in my head for years. A website. A blog engine. Astrological calculations. Consultation booking. Social media presence. Analytics. A brand that meant something to me beyond quarterly OKRs.
I knew the architecture. I knew the tech stack. I knew the content strategy. I could see it — fully formed, alive, running — every time I closed my eyes.
But when would I build it?
My days belonged to the company. Back-to-back meetings from 9 to 6. My evenings belonged to the code the team couldn’t finish. My nights belonged to production incidents and deployment windows. And whatever was left — those thin, fragile hours between midnight and sleep — those belonged to my family. To my kids. To being present at birthday parties where I smiled for photos while mentally debugging a payment flow in the background.
I’d tell myself: “Next month. When the release is done. When the migration stabilises. When the team ramps up. Then I’ll start.”
Next month never came.
The release was always followed by another release. The migration was always followed by another migration. The team always needed one more sprint of hand-holding before they could run independently.
And VedicJivan sat in my head — perfectly designed, never built. A blueprint with no builder.
I tried once. Carved out a Saturday afternoon. Set up the project. Got halfway through the homepage. Then Monday happened. Then Tuesday’s production incident. Then Wednesday’s stakeholder escalation. By the following Saturday, I’d forgotten where I left off. The momentum was gone.
That’s what nobody tells you about side hustles when you’re an engineering lead. It’s not about skill. It’s not about money — though that was tight too, every euro accounted for, no budget for hiring freelancers to build your dream. It’s about the one resource no delegation framework, no productivity hack, no time management course can manufacture:
Uninterrupted hours with a clear mind.
I didn’t have them. I hadn’t had them in years.
So the dream stayed a dream. Through promotions. Through company changes. Through moving countries. Through kids growing up. Through watching other people launch the things they talked about while I was still talking about mine.
The fire never died. But it was running out of oxygen.
And then — on a random evening when I expected nothing — everything changed.
The Evening Everything Changed
It was a regular weekday evening. Kids asleep. House quiet. I had maybe two hours before exhaustion would win.
I’d been hearing about AI tools changing the way developers work. I’d tried autocomplete assistants before — Copilot, code generators. Clever, sure. They could finish my sentences. But I didn’t need someone to finish my sentences. I needed someone who could understand the whole picture alongside me and carry forward the thought process.
That evening, I opened Claude — an AI assistant I’d been curious about — and instead of asking for a code snippet, I did something I’d never done with a tool before.
I talked about VedicJivan.
Not the code. The dream. The brand. The audience. The content structure. The tech stack I wanted. The astrology calculations that needed specialized APIs. The SEO strategy. The analytics setup with Google Tag Manager and GA4. The Instagram presence. The Facebook cover that needed to match a spiritual aesthetic.
Everything that had lived in my head for years — I poured it into a conversation.
And then something happened that years of delegation, documentation, and Jira tickets never achieved.
The AI didn’t just understand the pieces. It understood how they connected. It helped me architect the platform — not just the code, but the thinking. Content strategy for the blog. Node.js implementation for astrological calculations using specialised APIs. Brand voice options for Instagram. Analytics configuration to track which articles resonated.
I challenged it. “What about the mobile experience?” Addressed. “How do I structure the content for SEO?” Strategy outlined. “What’s the best way to handle consultation booking without a backend team?” Solution designed.
I wasn’t delegating. I wasn’t waiting. I wasn’t writing Jira tickets for someone else to interpret.
I was building.
For the first time in years — I was the architect AND the builder. With a thought partner who lost nothing in translation. Every requirement I described came back fully understood. Every nuance I mentioned was carried forward. The information didn’t leak. The context didn’t fade. The full picture in my head was finally materialising — exactly as I saw it.
In the span of a few weeks — working only in evenings, the same evenings that used to belong to corporate debugging — VedicJivan was alive. Website live. Blog publishing. Social media connected. Analytics tracking every visit. Astrological calculations running. A brand that looked like a small team of five built it.
One person built it. One person and one conversation.
I sat back and looked at what I’d created. The platform I’d dreamed about through years of birthday parties and production incidents and midnight deploys. The thing I’d told myself I’d build “next month” for half a decade.
It was real. It was running. It was mine.
And I hadn’t lost a single night of sleep to build it.
Here’s what actually changed — and it’s not what you think.
AI didn’t make me a better coder. I’ve been coding for fifteen years. It made me independent. The complete picture — functional, non-functional, security, maintainability, coding standards — finally had an execution partner that works at the speed of thought. No delegation. No information loss. No waiting.
I don’t depend on a junior developer’s availability to build what I can see. I don’t sacrifice my evenings to finish what meetings stole from my days. I don’t show up exhausted to morning standups because I was debugging at midnight.
And here’s the beautiful irony: that independence made me a better leader.
Because I sleep. Because I have energy. Because when a junior developer asks me to explain an architecture decision, I actually want to teach — instead of thinking about the five other things I still need to code tonight. Because the dream that was suffocating under corporate exhaustion is now breathing — and that changes how you show up for everything.
The lead who codes at midnight is a tragic hero in our industry. We celebrate them. We promote them. We burn them out.
I was that hero for fifteen years.
I don’t want to be anymore.
And thanks to AI — I don’t have to be.
The cursor still blinks. The screen still glows. But the coffee is warm, the hour is reasonable, and for the first time in my career — the night shift is over.
But lately, something else has been keeping me awake — and it’s not a production bug.
But What If the Cure Is Also the Poison?
I catch myself reaching for AI before I’ve finished thinking. I notice my patience for manual problem-solving shrinking. I wonder if my brain is learning new patterns — or forgetting how to form them on its own.
When a tool thinks alongside you at the speed of light, what happens to the parts of your mind that were built for slow, difficult, solitary thought? When the struggle disappears, does something else disappear with it?
The generation entering software engineering today will never know the 3 AM loneliness of a bug that doesn’t Google well. They’ll never build the mental calluses that come from hours of frustrated, unaided exploration. They’ll be faster. They’ll be more productive.
But will they be deeper?
I freed myself from the chain of delegation. But am I building a new chain — one made of prompts and convenience — that I can’t yet see?
The night shift is over. But the next conversation is just beginning.
That story — the one about what AI is quietly doing to the engineering mind — is one I’m still living. And I’ll share it when I understand it better.
→ Follow me to read Part 2: nandishdave.world
Nandish Dave is an Engineering Lead with 15+ years of experience building enterprise-scale digital platforms. Currently at PostNL, previously at bpostgroup. He writes about engineering leadership, AI-augmented development, and the human side of technology.