Tech

What sits behind the screenshots — the stack, how each thing is built, a robot coming together part by part, and the rooms where the hackathons happened.

stack

languages

TypeScript, C++, Python, JavaScript, SQL

frameworks

Next.js, React, tRPC, Motion

systems & tools

Supabase, Pusher, Vercel, Render, Git

focus

competitive programming, AI / ML systems, product design

Building education systems at the intersection of AI, product, and competition.

how it’s built

infrastructure

Studious

Next.js frontend on a tRPC backend; Pusher sockets for realtime, Supabase read replicas, load balancing on Render and Vercel at the edge.

→

ML · audio

notate

A decoder-only Transformer (6 × 512, 8 heads, ~21M parameters) trained on REMI-tokenised MIDI, played through a polyphonic synth written as a Web Audio AudioWorklet.

→

AI · hardware

Foundry

A project-aware copilot with streaming tool execution; mechanical CAD in KCL, schematic and PCB workspaces, a Monaco editor with Yjs multiplayer, Shopify checkout.

→

speech · vision

Synapse

Emotion read from voice tone and facial expression, attached to each message, and played back as expressive speech in the sender’s cloned voice.

→

AI · full-stack

Scribe

Frontend redesign, backend infrastructure, and the inference design behind its AI study workflows.

→

graphs

Nomad

Course planning on prerequisite dependency graphs, with a social layer for finding classmates and communities.

→

the robot

VEX team 15520X

engineer and programmer — 15520X (MECH_IT_HAPPEN)

3rd place skills and 3rd place qualifiers at Alpine Robo Games 2026.
Ranked 4th Skills, 7th Overall at Swiss Regional Competition (Santa Clash @ ISBasel).
Assisted in robot creation and design.
Engineered autonomous control mode.
Programmed interface for human interactions (controller).

driver control