Arnie's portfolio

Hardest thing I’ve done

I worked full-time at Apple, completed 28 credits at ETH Zürich (~a full semester), went to gym 4× a week and organized the Zurich OpenAI Robotics Hack (where I fundraised ~$30K in commitments in 1 week). This required working 12-14 hours every single day for 3 months.

Current project

CS:GO world models. I worked on robotics in Oct 2025; it confused me why VLAs use a VLM backbone. Thus, I started working on world models after my exams. Iterating on robots is very slow (requires real-world deployment), so I use CS:GO as my environment. I’m scraping game state (.dem) from FaceIt Pro matches, rendering gameplay, and training action-conditioned video diffusion / inverse dynamics / V-JEPA-style models.

Updates on my X.

Past technical projects

Humanoid in a week · Demo

With my friend Binh, we were the first team in the world to deploy TWIST2 full-body teleop, fine-tuned GR00T N1.5 on full-body data and ran low-latency inference to Unitree G1 (which we got through a LinkedIn post). Learnt how far behind open-source is, how bad teleop is, how poorly VLAs generalize, how bad deployment testing is.
GPT‑2 from scratch · Code

Coded up & trained on a 8xA100 cluster following Karpathy’s tutorial during exam revisions. Much more interesting than exam prep :)

Past experiences

Apple — Software Engineer Intern (Cork, Ireland) Mar 2025 – May 2025

Built an HTML editor for product managers to localize software update information. Integrated LLMs to automate the localization workflow. During n&w, I completed 28 credits at ETH Zürich, organized the Zurich Robotics Hack with OpenAI and went to gym 4×/week.
Amazon Web Services — Software Engineer Intern (Dublin) Jun 2024 – Sep 2024

Shipped a metric ranking the network risk of every AWS network device. Deployed 3 services each ingesting 1M requests/day; designed to be highly scalable using DynamoDB, Lambda, SQS/SNS, batching, etc.
Sleepiz — iOS App Developer Jul 2018 – Aug 2018

At 16, built their iOS app in 2 months. Login/auth, streaming biometrics; helped them raise 100K+ CHF in grants.

Some research work

Efficient, distributed sparse matrix–matrix multiplication · Report

Implemented a distributed SpGEMM algorithm in C++ & MPI; used primitives like Allgather(v), Alltoall(v), etc. Came up with a bloom-filter protocol to minimize column-sharing communication, leading to 33% speed-up (and beating Eigen in some cases).
Optimized Clifford neural networks · Report

Used kernel-caching, AVX2 SIMD vectorization, fused multiply-adds and multi-accumulator designs for ~5.5× speed-up over baseline for the linear layer; also implemented a kernel-free approach to reduce memory requirements further.

Past ecosystem projects

Zurich builds · OpenAI Robotics Hack

I wanted to surround myself with extremely talented, ambitious people (the current events in Switzerland sucked). So I organized Europe’s most talent-dense robotics hack, bringing together researchers (Lucas Beyer, Marc Pollefeys, Marco Hutter), frontier companies (OpenAI, Mimic Robotics, Loki Robotics, Prime Intellect) and exceptional builders (ex-NASA, ex-NVIDIA, ex-DeepMind, ...). Followed up with a simulation hackathon with Nvidia, Flexion Robotics and Jua.ai.
Ambitious x Driven · Site

The biggest challenge in Europe is ambition. So I launched a podcast interviewing exceptional students from ETH Zürich & EPFL to share their experiences and normalize ambition to ~1.8K subscribers. Closed a $2.5K partnership from Founderful Campus. I automated blog-writing using LLMs, then stopped posting to focus fully on building.
AI Tinkerers Zurich

Organized the first editions of AI Tinkerers in Zurich (~400 participants from DeepMind, Nvidia, Extropic, NASA, Isomorphic Labs, ...). Stopped due to bureaucracy and because the vibe drifted too much toward “networking” vs optimistic builders.

Achievements

Hummingbird Magnificent Grantee 2025 · Magnificent Grants

1 of 10 grantees selected in 2025.
FR8 cohort 1.0 · FR8

1 of 30 exceptional technical builders selected to spend summer in a Finnish hacker hotel. Worked on real-time human diffusion models.

Other stuff

more about me

Started coding at 12; at 14 built a Flappy Bird parody in Unity; at 15 built Android & iOS apps used by half my high school (~500 students).
At 15 launched a clothing enterprise at my school: sold ~1.3K clothes and made €20K revenue.
Organized the annual school festival as the youngest general manager (~5,000 attendees); youngest vice-president and president of my school student council.
Grew a following of ~20K on LinkedIn; quit in Feb 2026 after realizing most people actually building the future are on X.

Project backlog — projects I’d like to build eventually.

Real-time digital avatars

Collect 3D data using two iPhones with genlock; fit a Gaussian Avatar (SMPL-X + Gaussians on vertices); train a diffusion transformer conditioned on audio for face + body motion. I found digital avatars a bit dystopian, so I didn’t proceed (but the tech is very cool).
Large-scale 3D Gaussian Splatting

Early idea: scale hierarchical 3DGS with a retrieval-optimized “3DGS database” enabling out-of-core streaming (world-scale 3DGS, multi-client serving). DJI & other researchers beat me to it, so it became less interesting to invest time on.