Engineering notes — things I've shipped, things I've learned.
Long-form writeups on building production dashboards, scraping infrastructure, and the small architectural choices that decide whether a tool actually gets used.
My 12-step onboarding doc for every freelance project
Every freelance dispute I've heard about traces back to something that should have been agreed in week one and wasn't. This is the document I fill in with every new client during the discovery week. 12 items, ~30 minutes, zero post-kickoff surprises.
Read article →
MongoDB TTL indexes — auto-expire in one line, no cron
NeoMail expires old messages automatically with no cron, no scheduler, no cleanup code. The entire feature is one MongoDB index. When TTL indexes are the right tool, when they aren't, and the gotchas that catch most teams.
Read article →
The JWT + HTTP-only cookie auth I reuse on every project
Why localStorage JWTs are an XSS disaster, and the exact HTTP-only cookie + refresh token rotation setup I copy-paste into every React + Node project. Originally built for NeoMail, now in every dashboard since.
Read article →
The audit log that won finance teams' trust
A bare audit_log table, three rules for what goes in it, and a React side-panel UI. The single feature that turned skepticism into adoption on month-end close — same product, same UX, same charts, one feature decided it.
Read article →
The PostgreSQL window function that made rank history 50× faster
The query was 8 seconds. I needed it under 1. Replacing a self-join with a single LAG() window function — and changing the underlying table to suit it — got me to 0.15s. The schema decision behind the Amazon Analytics dashboard.
Read article →
Why my Puppeteer scraper kept getting blocked — and the 5 fixes that worked
Captcha-locked on Amazon Seller Central after two runs. Stealth plugin, persistent profile, real User-Agent, human timing, and captcha detection — the five fixes that finally made it reliable every Monday. With code.
Read article →
How I built an Amazon Ads analytics dashboard with Puppeteer Stealth + PostgreSQL — and cut weekly reporting from 4 hours to 5 minutes
A team I worked with sold on Amazon India. Every Monday: 3–4 hours producing a single ACOS/ROAS report. I replaced the pipeline with Node + Puppeteer Stealth + Postgres. Here's the architecture, the part that broke, and the schema trick that made rank-history queries trivial.
Read article →
I built 4 finance dashboards in 9 months — the React + Node playbook that worked every time
Different domains (reconciliation, AR aging, P&L, expense automation), same five underlying patterns. The denormalised reporting view, the freshness badge, the CSV export non-negotiable, and the audit log that won the team's trust during month-end close.
Read article →Want to chat about a finance dashboard, scraping job, or just say hello? Gauravthakur1k@gmail.com · LinkedIn · GitHub