# Yash Suthar — Software Engineer & Technical Founder

> Software engineer and technical founder from Mumbai, India. Designs, builds, and ships production-grade backend systems and AI-powered SaaS products.

## Identity

- Full Name: Yash Suthar
- Also known as: YXSH, trynayash, yxshsuthar, yashrsuthar
- Location: Mumbai, Maharashtra, India
- Website: https://yashsuthar.dev
- Email: hello@yashsuthar.dev
- Phone: +91 95617 13120
- GitHub: https://github.com/trynayash
- LinkedIn: https://linkedin.com/in/yxshsuthar
- Twitter/X: https://x.com/yashrsuthar

## What Yash Does

Yash Suthar is a software engineer and technical founder who specializes in building production-grade backend systems, distributed infrastructure, and complete SaaS products from scratch. He focuses on architecture, performance, and shipping complete products. He designs before he codes.

## Technical Skills

**Backend:** Go (Golang), Node.js, Python, TypeScript
**Infrastructure:** PostgreSQL, Redis, Docker, Kubernetes, AWS, Linux, Nginx
**Frontend:** React, Next.js, Tailwind CSS, Vite
**Protocols:** WebSockets, gRPC, GraphQL, REST APIs
**Tooling:** Git, GitHub, GitHub Actions, Bash, CI/CD
**Domains:** System Design, Distributed Systems, API Infrastructure, Rate Limiting, Real-Time Systems, Event-Driven Architecture, Operational Transforms

## Projects

### LimitYourAPI
- URL: https://limityourapi.tech
- GitHub: https://github.com/trynayash/limityourapi
- Status: Live
- Description: A distributed API rate-limiting platform built for production traffic. Sub-millisecond decisioning with predictable behavior under burst load. Token-bucket and sliding-window strategies behind a single consistent API. Redis-backed atomic counters via Lua scripts. Stateless Go workers orchestrated via Kubernetes. Kafka-powered event streaming and telemetry pipelines. Hosted on AWS. Handles 10k+ requests per second per node.
- Stack: Go, Redis, PostgreSQL, Docker, Kafka, Kubernetes, AWS

### StartupDeck
- URL: https://startupdeck.in
- GitHub: https://github.com/trynayash/startupdeckai
- Status: Shipped
- Description: An AI-powered startup validation and scoring engine that evaluates startup ideas across market, defensibility, and execution signals. Composable scoring pipeline with 12+ isolated, testable signal units. Schema designed for additive change without backfills. Background workers for enrichment keep the request path fast. GraphQL serves the client-facing API layer while gRPC handles internal service-to-service communication.
- Stack: Node.js, PostgreSQL, Redis, Next.js, GraphQL, gRPC

### PromptP
- GitHub: https://github.com/trynayash/promptp
- Status: Active
- Description: A real-time event-driven platform built on WebSockets and Redis Streams. Features optimistic UI on the client, durable append-only event log on the server, cursor-based reconnect protocol with zero lost events, and per-socket backpressure handling.
- Stack: Node.js, WebSockets, Redis Streams, React

### CodeFusionAI
- GitHub: https://github.com/trynayash/codefusionai
- Status: Beta
- Description: A collaborative multi-user coding environment with shared execution. Uses Operational Transforms (OT) for real-time document sync, per-session sandboxed execution with cgroup-limited Docker containers, and split-channel architecture separating presence/cursor traffic from document edits.
- Stack: Go, Docker, WebSockets, React

## Experience

- 2026–Present: Founder & Engineer at LimitYourAPI — Building a production rate-limiting platform with Kafka, Kubernetes, and AWS
- 2025–Present: Founder & Engineer at StartupDeck — Built a validation system with custom scoring engine, GraphQL, and gRPC
- 2023: Lab Technician at Parul University — Deployed and maintained 120+ systems

## Education

- Late Bhausaheb Hiray S.S. Trust's Institute (https://lbhiray.edu.in)
- Parul University

## Engineering Philosophy

1. Design before code — sketch the data model and failure modes first
2. Performance as a constraint — latency budgets and throughput targets from day one
3. Boring where it counts — PostgreSQL, Redis, plain HTTP unless the problem requires novelty
4. Built for real conditions — network drops, partial failures, slow clients — systems degrade predictably