/writing
What I'm writing about: agentic coding inside real teams, production LLM systems, AdTech AI, and dispatches from Bhilai. Newsletter weekly. Long-form when I've shipped something worth a writeup.
4 posts
First-Mile Founders, weekly on LinkedIn
- 2026 · 04 · 22Most MVPs die from scope, not competition
Why scope creep kills more zero-to-one products than any rival, and what to cut first.
- 2026 · 04 · 14Founder Voice vs. Brand Voice - where first-mile founders should spend their next hour
The two voices a founder writes in, why the first matters more early on, and how to tell which one is showing up.
- 2026 · 04 · 07Idea-Market Fit: 3 validation loops anyone can run in 48 hours
Three small loops to test demand before writing a line of code. Run them in a weekend.
- 2026 · 03 · 31Tech Stack in a Box: picking the right 'speed vs. control' mix
How to choose a stack when you have to ship fast but want to stay flexible. The trade-offs no template page tells you about.
8 posts
Technical deep-dives - AI systems, MCP, programmatic, infrastructure
- 2026 · 03 · 21Designing an event-sourced state machine for real-world fleet operations
Why the rental row never had a status column, and what we built instead.
- 2026 · 03 · 15The two-ledger pattern: separating money in from money out
Combined transaction tables look right on day one and are unrecoverable on day three hundred. The schema choice that makes audits a query.
- 2026 · 03 · 08Building a multi-source financial state model: reconciling CRM, payments, and bank into one view
How the console became a system of display, not a system of record - and why that decision held up under load.
- 2026 · 03 · 01Building a video-call stack that fails gracefully - from video to audio to record
Capture the consultation as data, not as the call. The discipline behind a telemedicine app that worked when the network didn't.
- 2026 · 02 · 22Daily route generation under real-world constraints: a hand-rolled scheduler
Why we abandoned OR-Tools for a deterministic algorithm operators could read. Trust beats optimality for local services.
- 2026 · 02 · 15Streaming under flaky bandwidth: bitrate adaptation, audio fallback, and reconnection logic
Buy the transcoding stack. Build the reliability layer above it. Make audio-only first-class. The notification is the product.
- 2026 · 02 · 08Bracket as a deterministic match graph: a fixture engine that supports manual override
Brackets aren't a rendering problem. They're a state-and-event problem with a particularly visual presentation.
- 2026 · 02 · 01Pessimistic vs optimistic concurrency for slot reservation: a benchmark and a decision
The standard 'use OCC' advice fits the median case. Under sustained contention with users who've already entered payment details, the honest choice is different.
8 posts
Long-form writeups of shipped products
- 2026 · 03 · 20Building an EV Rental Operating System for Last-Mile Delivery
Why the unit of billing was 'active rental day,' not 'monthly subscription' - and how that single choice rewrote the architecture.
- 2026 · 03 · 14Building a Recurring Giving Platform for Social Impact
Recurring donors don't churn loudly. They quietly stop trusting the platform. Here's how the product fought that decay.
- 2026 · 03 · 07Building a Financial Operations Console for MSME Businesses
Most B2B SaaS for small businesses fails the same way: it treats the owner like an analyst when they need an executor.
- 2026 · 02 · 28Launching a Telemedicine Platform During a Healthcare Crisis
Speed without breaking trust. The operating constraint when a healthcare app has to ship in days, not quarters.
- 2026 · 02 · 21Building a Subscription Commerce Platform for On-Demand Car Washing
Most marketplaces eventually become operations platforms. This one started as one - and that was the right call.
- 2026 · 02 · 14Bringing Local Temple Events to a Global Audience Through Live Streaming
Designing for an audience the tech industry doesn't usually design for - and what changes when you do.
- 2026 · 02 · 07Building a Tournament Management Platform for Local Sports Communities
Local tournaments don't need ESPN. They need the WhatsApp group of 80 organizers and players to stop arguing about the bracket.
- 2026 · 01 · 31Building a Multi-Location Turf Booking Platform Across India
Booking systems look easy from the outside. The hard part is what happens between the booking confirmation and the player arriving.
10 posts
General + life - founder reflections, notes, dispatches
- 2026 · 05 · 06My Claude Code session went great - until turn 30. What broke?
The first 20 turns felt like magic. The next 10 felt like a different model. Here's what was actually happening to my context window.
- 2026 · 05 · 06I asked for one thing. The agent did three. Where does the harness draw the line?
The agent finishes the task, then keeps going - refactoring three files I didn't ask about. The fix wasn't in the prompt. It was in the harness.
- 2026 · 05 · 06The agent hallucinated a tool argument. Whose fault was that?
I blamed the model. Then I read my own tool's JSON schema. It was the bug.
- 2026 · 05 · 06Cognition says no multi-agent. Anthropic ships sub-agents in Claude Code. Who's right?
Two of the sharpest teams in the field disagree out loud. After trying both, here's what I'd actually code on Monday morning.
- 2026 · 05 · 06Skill, slash command, sub-agent, or just a better prompt?
Four ways to codify the same recurring thing. I picked wrong, twice. Here's the decision tree I wish I'd had.
- 2026 · 05 · 06What does it mean for the harness to "trust" the model?
Every hook I write is a vote of no-confidence in the model. The question is which votes pay off.
- 2026 · 05 · 06My agent remembers me across sessions. I'm not sure I want it to.
I deleted one memory file and the next session got better, not worse. That bothered me enough to write this.
- 2026 · 05 · 06Why does Claude Code feel different from Cursor when they use the same model?
Same model, different feeling. Same task, different brain. The harness isn't infrastructure. It's the product.
- 2026 · 05 · 06I built my own agent harness over a weekend. Here's what I got wrong.
A while-loop, an LLM, and a tool registry. What could go wrong? Most of it.
- 2026 · 05 · 06Are skills the new programming primitive - or just better prompts in a folder?
The hopeful read: we just invented a new abstraction. The cynical read: we just invented imports. The honest read is in between.