LLM Caching Gateway
A four-layer cache in front of the model, proven by an eval harness and deployed on AWS ECS Fargate.
I'm Shivang, a full-stack engineer at Kfin Technologies. APIs, interfaces and LLM systems.
Latest buildLucida replaced 36 s AI drawing with 0.3 s search
Type an outfit, get real products in 0.3 s. Scroll to race it against my first version, which drew the outfit with AI first.Here it races my first version, which drew the outfit with AI first.
You ask: one request goes to both versions
0 drawing0 searches
A four-layer cache in front of the model, proven by an eval harness and deployed on AWS ECS Fargate.
Core platform work at Kfin: 30+ REST APIs and the services behind them.
Ask a database questions in plain English.
Faster validation over a legacy MSSQL system.
A 3D guide that explains mutual funds out loud.
Outfits from a chat (Flipkart GRiD 5.0, Level 2). It grew into Lucida.
A live site for a healer's therapy, mentorship and workshops.</>
Lets people donate spare food to listed NGOs. Built for Smart India Hackathon 2022.</>College hackathons to LLM systems at Kfin, in 79 seconds.grab a seat, turn the sound on
4 km loops, my reset button. Dogs always welcome.
made a friend on the routeA rally is just request and response, loudly.
smash incoming
Voice features, but make it personal.
mic in hand
Redstone taught me systems before AWS did.
build in progress
Always. Cold, with everything.
fuelLooking for full-stack and AI/LLM engineering roles.
thanks for walking the whole routePrefer it plain? The one-page version →