
Franklin Jara
I build type-safe web apps, product experiments, and small systems that earn their keep. Most of my work lives somewhere between production engineering, open-source contributions, and side quests that start from a real itch.
Lately I'm focused on shipping practical tools with AI where it helps, learning lower-level systems with Zig and Rust, contributing to projects like litter, and turning hobbies like bouldering into software. When I'm offline, I'm probably climbing or listening to techno.
build / ship / refine
AI products + R&D systems
Production Systems
Public Work Ledger
A compact index of products, templates, systems projects, and upstream contributions you can click through.
What shipped, not where I network.
The page now acts like a public shipping ledger: live products, starter kits, systems projects, and upstream pull requests with links back to the work.
Cima
AI-assisted climbing feedback, session history, grade progress, leaderboards, and 50+ badge loops for climbers.
Drezzi
AI outfit try-on with uploaded garments, personal photos, fast render loops, lookbooks, and 4K exports.
Mono-F7
A Turborepo template for TanStack Start, Hono, tRPC, Better Auth, Drizzle/Postgres, Docker, and deployable app scaffolding.
Zig builds
A programming language, HTTP server, POSIX-ish shell, tiny racing game, and Advent of Code work shipped in Zig.
AI Products + Distributed Systems
Cima
Cima turns climbing videos into structured feedback: technique notes, session history, grade progress, leaderboards, and 50+ unlockable badges that keep climbers coming back.
Drezzi
Drezzi renders uploaded outfits on your own photo in about 25 seconds, with 4K results, lookbooks, style tips, and multi-pose views built around a fast try-on loop.
Mono-F7
Mono-F7 is the starter I wanted for shipping full-stack ideas fast: TanStack Start, Hono, tRPC, Better Auth, Drizzle/Postgres, Docker, and Turborepo wired together.
Z-Language
A custom programming language written in Zig, from lexer and parser through code generation, built to understand the full pipeline behind fast developer tools.
[LABS]
F7
A multilingual full-stack todo app used as a proving ground for TanStack Start, React 19, tRPC, optimistic updates, and AWS/SST serverless deployments.
Open Source
I ship upstream PRs across Litter, Alleycat, Kittyfarm, sandbox-agent, Tokscale, Avante, and other tools I actually use.
Starlit AI
A generative art platform for creating and sharing DALL-E 3 images, built with React and AWS Lambda to explore prompt UX and serverless AI workflows.
Tiny Car
A tiny 2D racing game in Zig/Raylib focused on fast iteration: physics-based driving, random tracks, and real-time collision detection.
HTTP Server
A from-scratch HTTP/1.1 server in Zig that handles concurrent connections, routing, static files, and structured error logging without hiding behind a framework.
Generalist operating range
I like owning the seams: product shape, UI details, API contracts, queues, data, infrastructure, and enough systems work to know when an abstraction is leaking.
Ambiguous idea → useful workflow
React, TanStack, UX polish
Bun/Node, APIs, auth, queues
Postgres, Drizzle, SQL modeling
LLM workflows, agents, eval loops
AWS, SST, Docker, Turborepo
Rust/Zig when edges matter
Neovim, CLIs, dev environments