Ruchit Suthar
AI writes the code. I still make the call.
Fifteen years in, the question I ask about my work has changed three times. First it was “did it ship.” Then “is it right.” Now it's “should we build it at all.” I'm a software architect — I own the tech decisions, the trade-offs, and the ADRs for products where a wrong call costs real money and real users. I write about the call that's left after AI writes the code.
Ruchit Suthar
Senior Software Architect & Technical Leader
linkedin.com/in/ruchit07
Prefer email? Reach me at hello@ruchitsuthar.com.
The Climb
Output, then Quality, then Judgment
Three jobs, fifteen years. Each one added a layer without erasing the last. Here's the actual climb — including the parts that cost me.
- 01OutputFirst job · ~3 years
Engineer → Team Lead
Travel domain · a 15-year-old legacy product
My first real job was a fifteen-year-old travel product taking 1,500 requests a minute at peak. Before I was any good at writing code, I was good at keeping it running — server setup, networking, security config, reading logs late at night, finding the bug nobody else could. Then I built the things the business ran on. The rule here was simple, and I believed all of it: output matters, whatever it takes.
- London Pass QR-code validation
- Expedia & Viator supplier integrations
- Led a team of 5 at ~3 years in
What it taught meI learned the system from the metal up — when caching helps, when to keep data in memory, what breaks at scale before it breaks. You think like an architect early when you've already been the one woken up to fix it.
What it cost meI made every hiring call here. Once I said yes to a candidate for a sad story, not the skill in front of me. The team carried that gap, not me. A sad story is not a hiring signal — I learned that the hard way.
- 02OwnershipSecond job · ~3 years
Full-SDLC IC → Mentor
10-person startup · US clients
At a ten-person startup I became a full-SDLC IC — took requirements from clients, built, tested, shipped, solo. Then the team grew and I led and mentored the juniors. That was the start of my mentorship journey. By six years in, I could build and launch a product on my own, make the trade-offs, and own the whole delivery.
- Org-survey tool used by Coca-Cola & Pfizer
- High traffic, large datasets, mass emailing
- Owned delivery end to end, solo to team
What it taught meOwning delivery end to end taught me where design decisions actually get made — and that mentoring people is leverage, not overhead. Build it, launch it, answer for it.
What it cost meAs the most senior person I ran every interview and made the tech hiring calls. That's where the cost of emotional hiring — saying yes to the story instead of the signal — stopped being abstract.
- 03JudgmentNowCurrent · many years
Software Architect & R&D
Fortune 500 · engineering-led, quality-obsessed
Then a Fortune 500, and work I used to ship in a week took a month. Not because anyone was slow — there was a real process at every step, and for the first time an architect reviewed my code before production. I didn't like it. Then I understood it. Clean code, design patterns, ADRs. That lead is the reason I became an architect. Now I own the call: tech stack, infrastructure, the trade-offs nobody else wants to sign off on — including what AI should and shouldn't be trusted to do.
- Custom agent orchestrator — MVPs 3× faster
- AI in real products: prompt → context → MCP → RAG → agentic
- Leads a team averaging ~8 years — every one a personal hire
What it taught meAI didn't replace judgment. It moved it. The rare skill is no longer writing the code — it's deciding what's worth building, then checking what the machine gives back. The new bottleneck isn't speed. It's review.
What it cost meA Lambda I left running in a loop quietly burned $5,350 in a month. A product that missed the quality bar cost three months of rework. And once I stayed too nice with someone who needed the truth, until they ran out of time. The kindest thing a leader can do is say it early, while there's still time to fix it.
How I Operate
Survival patterns, not values
Technology humans can maintain. Systems teams can understand. Decisions that compound positively over years. Fifteen years of production systems taught me each of these the hard way.
Production is the proof
Every architectural decision gets one test: what breaks at 3am, and who gets paged? If the team can't maintain it without the person who built it, it's not architecture. It's a liability with a deploy date.
Constraints over patterns
Knowing microservices from monoliths matters less than knowing which constraints are real and which are assumed. The right architecture fits your actual team, budget, and timeline — not the case study.
AI is infrastructure now
Not a feature, not a team, not a strategy doc. AI belongs in the system-design conversation from day one — the same as databases, caches, and queues. Treat it as an add-on and you'll refactor it out in two years.
Hiring is architecture
1,000+ interviews taught me that team design is system design. Who you hire, how you onboard them, what problems you let them own — the most leveraged architectural decisions you make. Gut feelings don't scale.
Sustainable beats heroic
Remote-first across five time zones taught me this fast: sustainable pace beats hero culture, every time. A system that needs heroes to operate is one incident away from failure. Design for median capacity, not peak heroism.
Writing forces clarity
If you can't write down why you made an architectural decision, you probably don't fully understand it yet. I write to verify my own thinking — and to pass it on in a form that survives the next reorg.
Beyond the Work
The same standard, off the clock
The operating system doesn't switch off after work. Swiss watches and rare-find sneakers for the craft — I don't follow the trend, I build my own. A national-level volleyball habit for the discipline. A wishlist I wrote in 2018 and finished in 2025, then spent three hours on a Milan-to-Venice train writing the next one. And sunrises I wake at 5am to watch from the best vantage, phone down, just watching.
It's the same temperament that makes a good architect: patience, attention, presence. Quality you can feel is the whole point — in a system or anywhere else.
What I'm Writing About