Builder / Founder

I build and ship production-ready products.

10 products shipped. Multiple live deployments. 2-week MVP cycles.

No endless planning. No theory. Just working code deployed to production.

LAST LAUNCH 5 days ago
CURRENT STACK Next.js + Supabase

Track Record

5
Projects Shipped
From idea to production
14 days
Avg MVP Time
MVP to live
20+
Deployments
Across all products
5+
Tech Stacks
Next.js, Java, Node

Shipped Products

Rediate
LIVE FOUNDER
Next.js 15 • React 19 • Next.js API Routes

Problem

Creators waste hours scrolling Reddit and communities to find real signals. Even when they find something interesting, turning it into structured, publishable content is slow and manual.

Solution

Rediate analyzes subreddits and conversations, extracts real user pain points, and converts them into ready-to-use content ideas that can be refined into newsletters, videos, blog posts, or social threads.

Highlights

  • Asynchronous job system with real-time status polling
  • Multi-format refinement (newsletter, video, outline, auto-detect)
  • Cost-controlled beta using quotas and IP-based rate limiting
  • Approach: Event-driven async job pipeline: Subreddit → /api/start → Reddit fetch → LLM analysis → Job status polling → Results → Refinement → UI. Jobs run asynchronously with progress states (queued, analyzing, drafting, completed) so UI never blocks.
  • Scalability: Cost-controlled via API quotas and rate-limiting. Async job processing prevents blocking. Optional KV for rate-limit tracking.
  • Deployment: Vercel serverless deployment with Next.js API routes. Optional Firebase for persistence beyond KV.

Metrics

Growth: 1 analysis/day, 8 refinements/day
Performance: ~45s avg analysis | $0.10 cost

Timeline

Total: 3 months
Launched: Nov 4, 2025

Key Learnings

  • Raw data (Reddit) is more valuable than generic prompts
  • Users want refinement, not just generation
  • Signals > vibes when creating content ideas
Stack: Next.js 15, React 19, Tailwind CSS, Framer Motion, Next.js API Routes, OpenAI API, Reddit OAuth, Vercel, Optional KV / Firebase for quotas & storage
FairSplit
LIVE CO-FOUNDER
Next.js 15 • React 19

Problem

Splitting shared expenses in groups is messy: manual math, unclear "who owes whom", and friction when settling up.

Solution

FairSplit is a local-first web app to create expense rooms, record shared expenses, and automatically compute an optimized settlement that minimizes the number of transactions—no backend required.

Highlights

  • Local-first + offline-capable (no login, no backend)
  • Clear settlement summary (who owes whom)
  • Greedy algorithm that reduces settlement to fewer transactions
  • Responsive UI with MUI + animations
  • Approach: Client-only architecture: rooms + expenses managed in local state and persisted to Local Storage. Settlement is computed on-demand from current room state and rendered as a clear "who pays whom" summary.
  • Scalability: Optimized for small/medium groups (instant client-side computation). Zero server cost. Works offline; data survives refresh and sessions.
  • Deployment: Static Next.js deployment on Vercel. No server infrastructure needed.

Metrics

Growth: Greedy settlement algo
Performance: Zero backend cost

Timeline

Total: 14 days
Launched: Jul 22, 2025

Key Learnings

  • Reducing user friction beats adding complexity (no accounts, offline-first)
  • Showing a "simple settlement" output prevents disputes
  • Leading a co-founder team: delegation and clear responsibility separation drives velocity
Stack: Next.js 15, React 19, Material-UI, Framer Motion, Vercel, Local Storage
Sherlog
BETA FOUNDER
React • Tailwind CSS • Rust

Problem

Debugging production systems from raw logs is painful: millions of lines, noisy stack traces, duplicated errors, and no way to see what actually matters without manual grep, regex, and guesswork.

Solution

Sherlog is a privacy-first log analysis engine that ingests large log files, automatically groups similar errors, extracts root patterns, and generates meaningful summaries and regexes to help engineers understand what is really happening in production systems. Instead of reading logs, you analyze signal.

Highlights

  • Clusters millions of log lines into meaningful error groups
  • Automatically extracts regex-like patterns
  • Works entirely offline with 100% privacy
  • Built with Rust + WASM for speed
  • Approach: Log files are loaded directly into the browser, processed through a WASM-compiled Rust engine that performs parsing, tokenization, similarity grouping, and pattern extraction. Results are streamed back to the UI in real time.
  • Scalability: Handles GB-scale logs locally. Zero server bottlenecks. Performance bound only by the user's CPU.
  • Deployment: Browser-native WASM execution. No server required. All data stays on client.

Metrics

Growth: 100% local execution
Performance: Client-side, zero network latency

Timeline

Total: Ongoing experimental
Launched: Sep 1, 2024

Key Learnings

  • Developers don't want raw logs — they want structure
  • Privacy-first tooling is a real differentiator
  • Speed beats fancy UI in debugging tools
Stack: React, Tailwind CSS, Rust, WASM, Browser-native
Airing Calendar
LIVE FOUNDER
React • TypeScript

Problem

Tracking anime release schedules across multiple sites is messy, inconsistent, and full of spoilers. I wanted a single clean place where I could see exactly what airs today, without noise.

Solution

Airing Calendar is a lightweight web app that shows upcoming anime episodes in a clean, calendar-style view. It pulls real airing data from Jikan (MyAnimeList API) and presents it in a fast, minimal interface optimized for daily use. It's the tool I personally use to decide what to watch tonight.

Highlights

  • Real-time anime airing schedule
  • Clean daily view
  • Zero spoilers
  • Built for fast daily usage
  • Approach: Client-side app that fetches live airing schedules from Jikan and maps them into a date-based calendar UI.
  • Scalability: API-driven, stateless frontend. Can handle thousands of users with no backend.
  • Deployment: Static site with client-side API fetch from Jikan.

Metrics

Performance: API-driven, instant refresh

Timeline

Total: Ongoing
Launched: Aug 15, 2024

Key Learnings

  • A product doesn't need millions of users — it needs one user who cares
  • Solving your own pain produces better UX
  • Consuming third-party APIs at scale
Stack: React, TypeScript, Tailwind CSS, Static deployment
Personal Website 2.0
LIVE FOUNDER
Astro • TypeScript

Problem

Most developer portfolios are static resumes. They show skills, not execution. They fail to communicate velocity, product thinking, or real-world impact.

Solution

A brutalist, minimal, product-oriented website that showcases shipped products, timelines, metrics, and what was learned — not just screenshots. Designed to look more like a product dashboard than a personal site.

Highlights

  • Brutalist minimal UI focused on clarity and speed
  • Expandable product cards with full technical breakdowns
  • Live "currently building" section
  • Designed to evolve as a living product, not a one-off page
  • Approach: Static-first, ultra-fast delivery with dynamic data injected at build time. Focus on performance, clarity, and zero bloat.
  • Scalability: Pure static hosting. No server required. Instant global delivery.
  • Deployment: Vercel / Cloudflare static hosting with build-time data injection.

Metrics

Performance: Ultra-fast static delivery

Timeline

Total: 2 weeks
Launched: Jan 28, 2026

Key Learnings

  • Recruiters scan, they don't read
  • Track record > buzzwords
  • Brutalist UI increases signal-to-noise
Stack: Astro, TypeScript, Tailwind CSS, Vercel, Cloudflare

Currently Building

IN PROGRESS

AI Dispatcher

AI orchestration layer that automatically routes tasks to optimal agents based on complexity, cost, and capability.

Week 2 - Building real-time AI capacity management
STACK
Claude APIOpenAIGoCLI
Target Launch: Live by end of month