They said I can’t. UKANT did.
AI can ship. I make it earn.
I’m Ukant. I lead backend engineering at Caprae Capital, an M&A firm. I own the backend, the Stripe integration and the AWS infrastructure, and I lead the DB and Email teams. I build with AI every day, and my job is to turn vibe code into production code.
The numbers I’m willing to defend: API latency dropped from 2 seconds to 20 ms (100x faster). Docker images went from 1.5 GB to 220 MB (87% smaller). Both shipped, both still running.
Before Caprae, I built enterprise data pipelines at Mente Consultancies in SAS and SAS Viya for a banking client: Calendar and Rebase Marts, statistical variables, and dashboards that improved data quality by 25% and cut drift by 30%. Before that, an IEEE paper on cardiac ECG classification: six months of work, 80% accuracy and 90% ROC-AUC, the best of six models on the PTB-XL dataset.
I learn by breaking things and rebuilding them. Two years on a riced DWM, tmux and nvim setup will do that. I share what doesn’t work as readily as what does, and I’d rather ship something rough than wait until it’s perfect.
I write Steal Like an Engineer: a feature from a famous product, the trick behind it, and how my team used the same idea.
What I want is direct: meaningful work, freelance independence, and money. The kind of money that helps my family. Hiring is secondary. I’m building.
📍 Udaipur, India · IST · Remote-first
BTech in Computer Science and Engineering with AI & ML, 2024
Sir Padampat Singhania University
High School in Mathematics, Physics and Chemistry, 2020
Oasis Sainik School
What I use in production. Numbers attached where they matter.
Python (daily) · SQL · Dart · Bash · TypeScript (learning)
REST · Microservices · Stripe · SSO · 100x latency wins (2s→20ms)
DynamoDB · SQL (multi-tenant 3NF) · Redis · MySQL
AWS (EC2, Route53, DynamoDB, S3, ECR) · Docker (87% shrink) · CI/CD · Nginx · systemd
TensorFlow · Keras · MLFlow · scikit-learn · SAS Viya · IEEE-published
Hey! let’s build something together.