Backend Engineer at EPYC | Building distributed systems, compilers, and everything in between

I'm passionate about backend engineering, distributed systems, and systems programming. When I'm not shipping production code, I'm probably implementing a consensus algorithm or building a programming language for fun.

📍 Links: Blog • Email • Twitter • Resume

Languages            TypeScript · Go · Kotlin · JavaScript
Backend              Node.js · Cloudflare Workers · Hono · Express.js
Cloud & Infra        AWS (Batch, Lambda, S3) · Cloudflare · Pulumi
Databases            PostgreSQL
Frontend             React · Next.js · Remix · React Router · Vite

Backend Engineer @ EPYC | Current

Delivering production systems for clients across scheduling, fintech, and funding platforms. Here's what I've built:

Built a robust appointment booking platform connecting users with industry experts.

Highlights:

• Zero double bookings through proper concurrency control
• Integrated Razorpay with duplicate payment prevention
• Google Calendar sync to avoid conflicts
• Custom mentor availability management

TypeScript React Router v7 Cloudflare Workers PostgreSQL

Enterprise credits management system for airline employee uniform purchases.

Highlights:

• Balances never go negative - handled every concurrency edge case
• Accurate accounting across refunds, cancellations, and returns
• Custom tax software integration for precise sales tax
• Analytics dashboards for vendors and airline management

TypeScript Next.js Cloudflare Workers PostgreSQL

Collaboration platform connecting founders with investors.

Highlights:

• Application reconciliation across Algolia + multiple internal APIs
• Fault-tolerant architecture - graceful degradation when APIs fail
• Duplicate detection and missing application recovery

TypeScript React Router v7 Cloudflare Workers PostgreSQL

• Built bulk Slack DM system sending 10,000+ messages with rate limit handling (Bull queue)
• Decomposed monolith → microservices, slashing memory/CPU usage during high job loads
• Developed Stripe payment gateway for events platform
• Migrated services Digital Ocean → AWS, improving performance
• Created dev environment CLI: setup time from 2+ days → minutes
• Led JS → JSDoc migration: fewer runtime errors, zero delivery slowdown

I build things to understand how they work - from consensus algorithms to compilers.

Fast fuzzy string search using finite automata - 10,727 queries/sec on 370K-word dictionary.

Performance: 38% faster than runtime construction at distance 3
Approach:    Precomputed DFA + Trie-based dictionary
Language:    Kotlin

7 implementation versions (v1-v7) showing progressive optimizations from the Schulz & Mihov paper.

Raft consensus algorithm implementation with formal verification before a single line of code.

Verification: P-lang model checking
Approach:     Minimal mutex usage for performance
Language:     Go

Links: P-lang Models • Original Paper

Network programming challenges: TCP/UDP servers, custom protocols, and everything low-level.

7 Challenges Solved:

• Custom LRCP protocol (UDP + acks + retransmission)
• Binary TCP protocol for speed cameras
• TCP proxy with address rewriting
• Multi-user chat server
• Asset tracking with binary protocol + SQLite
• And more...

Go TCP/UDP Custom Protocols

Custom language with full compiler pipeline: Lexer → Parser → AST → Compiler → VM.

Features: Variables, arithmetic, boolean ops, REPL
Output:   Bytecode execution
Language: Go

Serverless distributed rendering: distribute animation frames across AWS Fargate containers.

Architecture: AWS Batch + Lambda + S3
Orchestration: S3 upload events → Lambda triggers
IaC:          Pulumi

Parallel rendering with automatic frame consolidation. Go Docker AWS

TypeScript → Native code using LLVM. Educational compiler with full test coverage.

Supports: Variables, control flow, functions, operators
Backend:  LLVM
Output:   Native code

Temporary email for developers - test OTPs, password resets, without inbox clutter.

Features: Real-time WebSocket updates • AI-powered code extraction (GPT-4o-mini) • 15-min auto-cleanup • Zero registration

Next.js 14 Cloudflare Workers Durable Objects D1

Real-time video broadcasting with XState state machines for complex UI flows.

Features: Multi-participant calls • Screen sharing • WebSocket signaling (Durable Objects) • State machine visualization

React Router v7 WebRTC XState v5 Cloudflare Calls API

Query JSON files with natural language. Your data never leaves your browser.

Features: GPT-4 powered • Auto schema generation • Streaming responses • Client-side processing

Remix Vercel AI SDK Cloudflare Pages

Kubernetes operator for PostgreSQL clusters - backups, PITR, on-demand DB provisioning for PR previews.

TypeScript Effect-TS Kubernetes PostgreSQL 17 pgBackRest

Completed 6 challenges: Echo • Unique ID Generation • Single/Multi-Node Broadcast • Fault-Tolerant Broadcast • Efficient Broadcast

When I want to learn a concept deeply, I prefer challenges with clear constraints over building full projects.

Path Finder • GitHub Pathfinding algorithm visualizer - one of my first projects.

slate-devtools • GitHub Debugging tools for Slate.js rich text editor. Built while cloning Notion as a beginner.