Skip to content
View sinhaankur's full-sized avatar
🌴
On vacation
🌴
On vacation

Block or report sinhaankur

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
sinhaankur/README.md

Hi, I'm Ankur 👋

Profile views

Principal UX Designer at Oracle · Human–AI Interaction · Toronto

I'm a designer who ships code. 12+ years designing enterprise products across banking, supply chain, and cloud infrastructure — and I don't stop at Figma: I research the problem, design the interaction, and prototype it in real code, then hand engineering a contract they can build against. At Oracle I own the design end-to-end for cloud database operations in OCI, turning dangerous, expert-only tasks (encryption, key rotation, disaster-recovery failover) into safe, guided flows a non-expert can run.

The open-source work here is where that thinking gets to be fully public. Most of these repos are exploration and ideas — a working sketchbook for the human–AI seam, motion, performance, and on-device AI. A few (like Unhosted) are serious builds. The day job at Oracle is the anchor; everything else is how I keep sharp.

🎨 How I work as a designer

  • Research → design → code prototype → handoff. I talk to real users (DBAs, operators), map the flows, and prove the interaction in working code before engineering commits to building it.
  • Reversibility, not "safety," is my policy axis. How I decide when an interface must pause for a human: the more irreversible the step, the more the UI slows you down and confirms.
  • Performance is a UX concern. A design isn't done if it janks. Example: for the Universe Engine on my site, I render body/constellation labels as in-scene WebGL sprites (so they occlude at true depth instead of floating over the 3D scene), and measure/wrap their text reflow-free with @chenglou/pretext — the canvas font engine does the layout, so building a label never forces a browser reflow, and multilingual names (CJK / Arabic / emoji) still wrap correctly.
  • Calibrated language over raw numbers. In my AI prototypes, confidence reads as Likely / Unsure / Low (exact % on hover) — because a percentage isn't a decision, a posture is.

🚀 Unhosted — AI that lives where you do

Frontier-class AI inference on hardware you own. Unhosted pools the computers you already have — and optionally your friends', and optionally a public swarm of strangers' GPUs — into a single inference cluster. One endpoint. Mac, Linux, Windows. CUDA, Metal, ROCm.

Three trust modes. You decide the radius:

       ╭───────────────────────────────╮
       │   public · pay (USDC)         │   strangers' GPUs, opt-in
       │   ╭───────────────────────╮   │
       │   │  trusted · free       │   │   friends, family, team
       │   │   ╭───────────────╮   │   │
       │   │   │ local · free  │   │   │   devices you own
       │   │   ╰───────────────╯   │   │
       │   ╰───────────────────────╯   │
       ╰───────────────────────────────╯

The first two are free forever. The third is the safety net. You can use Unhosted for the rest of your life and never spend a dollar.

Status: pre-alpha, built in public. CLI at v0.0.34.

Repo What it is Lang
unhosted-core The daemon — LAN cluster, mDNS discovery, model management, single-endpoint API. Rust
unhosted-payments Settlement layer for public mode — policy-driven, multi-rail, country-aware. Rust
unhosted-plugins Extensions that talk to the daemon. First: MCP server. TS
homebrew-unhosted Homebrew tap. Today: RPC-enabled llama.cpp for VRAM-pooling. Ruby

💼 Day job — the anchor

Principal UX Designer at Oracle, OCI Database Tools (DBaaS team) — I own the design end-to-end for cloud database operations: encryption (TDE + OCI Vault/KMS), pluggable-database lifecycle, and Cross-Region Data Guard failover. Turning dangerous, expert-only tasks into safe, guided flows a non-expert can run. Feature specifics are under NDA — the open-source work here is my best public demonstration of how I think about design.

🧪 Exploration & ideas

A sketchbook, not a product shelf. These are how I pressure-test the ideas above in real code — some polished, some deliberately rough. Grouped by what they're exploring:

Human–AI seam & on-device AI

  • cognitive-twin-agent — a local-first personal-AI runtime; the seam between a human and an agent, on hardware you own
  • EMPATHEIA — camera-based empathetic AI companion, offline-first hybrid fallback
  • Probabilistic-UI — a design vocabulary for AI surfaces that are honest about uncertainty

Engines & real-world data (the visual, performance-first lane)

  • portfolio-2026 — the Universe / Satellite Engine: real orbits, live space data, WebGL, sinhaankur.com
  • Firmament — point your iPhone at the real sky and understand it (iOS, Apple Intelligence)
  • MultiVerse — physics-grounded Three.js visualization of cosmological models
  • aero-engine-3d — interactive 3D encyclopedia of Airbus families

Apps & tools

  • Kelo — private, on-device health + wealth companion (Swift, HealthKit)
  • Draften — a unified design + diagramming workspace that opens your files (Tauri)
  • Structura — scan real structures with your phone's depth sensor (iOS/Android)
  • WatchTower — turn a computer you own into your personal cloud (self-hosted deploys)
  • GovLens — a universal reader overlaid on government portals across 25+ national TLDs

Explainers

Full map of the live projects: sinhaankur.com/lab

🌐 Find me

@sinhaankur's activity is private