Yash Garg
— Systems architect
I build systems that ship, interfaces that surprise, and ideas that shouldn't work — until they do.
── NEW FIELD NOTE / 2026.07.30
Five months
in the arena.
I joined Matiks on February 23. The next day, my first pull request was up. What I built, what I learned, what hurt, and what I am carrying forward.
Read the field note →
Stop scrolling like a normal person.
YASHIMAGATORU is live.
A cursed genius, a chessboard with god issues, and reality getting bullied by code. This is the next click. The portfolio can wait.
── 01 / What I Build
How I think about building.
I sit at the intersection of systems engineering, product thinking, and interface craft. I don't separate the technical from the creative — the best systems are both.
Systems Engineering
Rust streams. Caching layers. Cost engineering. I think in throughput, latency, and failure modes — then build systems that hold up under real conditions.
Product Architecture
BFF patterns. GraphQL migrations. Clean contracts. Shaping systems around user journeys — not around accidental technical sprawl or legacy convenience.
Interface Design
Tactile UIs. Meaningful motion. Delight in the details. Software that feels alive — that rewards exploration and makes you feel something without knowing why.
AI Products
Agentic interfaces. Voice-first apps. AI coworkers. I build the parts of the future that are technically possible right now but nobody has shipped yet.
Game Mechanics
Engagement loops. Reward systems. Interaction design with real stakes. Co-founded a game dev collective in college. Never really stopped thinking in levels.
Experiments
Weird ideas. Ambitious concepts. Things that shouldn't work but kind of do. The lab never closes. Half the best ideas look unreasonable until they ship.
── 02 / Selected Work
Missions, shipped.
Not just what I built — but what changed because of it. Click any dossier to read the full build story.
── 03 / Ideas Lab
The lab.
Where ideas live before they become products. Some of these are building. Some are concepts. Some are just too interesting to ignore. Hover to see the detail.
Voice-Driven App Builder
Talk to it. It builds. Not prototypes — actual deployable apps. The IDE as conversation, the computer as collaborator.
AI Coworker by Email
An AI colleague you CC on tasks. It picks things up, follows up, keeps context across threads. Async intelligence.
AI Teacher with a Blackboard
Not a chatbot. A visual thinker. It draws diagrams, writes proofs, solves problems with the physical metaphor of teaching.
Agentic Airline Interface
A conversational layer over airline e-commerce that actually knows your preferences, history, and intent.
Debate Rooms
A space where ideas get contested. AI-assisted argumentation, structured discourse, real intellectual stakes.
Social Feed of AI Chats
What if your feed was a conversation? Share an AI exchange. React with a counter-argument. Build on each other's thinking.
— the lab never closes —
── 04 / Principles
How I build.
Not generic values. Real builder principles — the ones that actually shape decisions when things get hard.
Software should feel alive.
Interactions should have weight. Transitions should have memory. If a UI doesn't make you feel anything, it's not finished yet. The emotional quality of software is part of its correctness.
Architecture should reduce chaos, not rename it.
The goal isn't to move complexity somewhere new. It's to eliminate it. A microservice that hides the wrong abstraction is worse than a monolith with a bad name. Good systems get simpler as they mature.
Interfaces should reward curiosity.
The best UIs have layers. Surface-level for the casual. Depth for the explorer. Design for the second visit, not just the first. The things you discover on your own are always more memorable.
Product quality includes emotional quality.
Users don't just evaluate features — they feel software. A system that works but feels bad is still failing. Performance is a feature. Delight is a feature. Personality is a feature.
Build weird things. Ground them in craft.
Weirdness is valuable when backed by real skill. The most interesting ideas sound unreasonable until they ship. The craft is what makes them land — what separates a toy from a tool.
── 05 / Right Now
What's occupying the stack.
Real-time status. What I'm working on, thinking about, and excited by right now.
$ last updated: July 2026 · open to ambitious problems
── 06 / Contact
Let's build
something.
If you're working on something ambitious, weird, or genuinely difficult to describe in a single sentence — I want to hear about it.
Smart people building hard things. That's the only filter.