Shipped, and stillrunning.

Audited in 48 hours. Fixed in four days.

Glimpse Markets shipped a crypto prediction market fast on AI-generated code, and inherited a slow, fragile trading frontend. The 48-hour audit ranked what actually mattered: security holes by severity, untyped boundaries, and a browser-side data fan-out that taxed every trader.

The first week of fixes landed four days after go-ahead. Market data now routes through their own backend, and first paint dropped from 692ms to 132ms. It runs today on a reliability retainer.

"Sushil found a week of latency hiding in our frontend and shipped the fixes in days. First paint went from sluggish to instant." — Ruban, founder · Glimpse Markets

AI writes your training plan. Code decides if it ships.

Leverage builds a full strength-training block from your goal, schedule, and equipment. Then deterministic checks gate every program before it renders: push and pull volume stay balanced, skill work sits first while you're fresh, and progressions advance only on held standards. An L-sit has to hold for 20 seconds before the next step unlocks.

Hand-written vanilla JavaScript on PocketBase. No build step, installs as a PWA. I train on it daily; the AI generation is useful because the rules that govern it are written in code.

Trading strategies, versioned like software.

Manja is a portfolio manager for NSE equities. It runs several trading strategies side by side, tracks how each one performs, and versions them like software so improvements ship as new releases.

A strategy only earns real weight after it proves itself on live results, not backtests: the same evaluation discipline I put around AI features before they ship anywhere.

Three more builds in motion.