About / Since 2019

Kami memulai dari pertanyaan
yang belum dijawab sistem modern.

Fantasma Synergy menyatukan applied AI, akselerasi hardware, dan arsitektur edge untuk membuat keputusan otomatis tetap dapat diamati, diuji, dan dipercaya.

2019: Runtime v1
2022: Edge Lattice
2025: Human Signal

2019 / The first runtime

Sistem tidak boleh menunggu koneksi untuk membuat keputusan.

Kami memulai dengan runtime yang dapat menjalankan model dekat dengan sensor dan aktuator. Target awalnya sederhana: membuat inferensi di lapangan cukup cepat untuk menjadi bagian dari keputusan, bukan laporan setelah kejadian.

2022 / The lattice

Edge adalah lingkungan utama, bukan versi kecil dari cloud.

Ketika deployment berkembang dari satu perangkat menjadi ribuan node, kami membangun Edge Lattice: jaringan yang dapat menyelaraskan konteks, bertahan saat link putus, dan tahu kapan data benar-benar perlu keluar dari lokasi.

2025 / Human signal

Bukti bukan dokumentasi tambahan. Bukti adalah fitur produk.

Setiap keputusan harus bisa ditelusuri ke sinyal, model, dan kondisi operasionalnya. Prinsip ini membuat tim kami dapat membawa Human Signal ke konteks sensitif tanpa menyerahkan privasi atau keterbacaan sistem.

Operating principles

Cara kami membangun.

Lima prinsip ini dipakai untuk memilih masalah, menulis RFC, dan menguji sistem sebelum ia masuk ke lingkungan nyata.

  1. 01

    Mulai dari batas fisik

    Kami merancang dari latensi, daya, konektivitas, dan konsekuensi kegagalan yang nyata.

  2. 02

    Sinyal sebelum antarmuka

    Sistem yang baik memahami konteks sebelum meminta pengguna melakukan pekerjaan tambahan.

  3. 03

    Observability adalah default

    Setiap keputusan penting menyisakan jejak yang cukup untuk diperiksa ulang.

  4. 04

    Privasi tinggal di perangkat

    Data sensitif diproses sedekat mungkin dengan sumbernya dan keluar hanya saat diperlukan.

  5. 05

    Kecepatan dengan alasan

    Kami bergerak cepat karena fondasi teknis, pengujian, dan ownership dibuat eksplisit.

The team / 140 people

Orang yang memegang sistem sampai tuntas.

Lihat karier terbuka
AM

Dr. Arjun Mehta

Co-founder & Chief Scientist

Arjun built the first version of Neural Fabric in a rented server room in Singapore in 2019. Previously: principal researcher at Google Brain (2014–2019), PhD in machine learning from ETH Zürich. He thinks in systems, writes in proofs, and runs 40km a week.

YT

Yuki Tanaka

Co-founder & CEO

Yuki spent seven years building distributed infrastructure at Stripe and Square before co-founding AETHER. She is obsessed with the gap between what systems promise and what they deliver. Headquartered in Singapore; often in Zurich.

SK

Dr. Selin Karaçay

VP Research

Selin joined AETHER from MIT CSAIL, where her PhD thesis on distribution shift in production ML systems became the foundation for Neural Fabric's drift detection engine. She leads a team of 18 researchers across two continents.

JO

James Okafor

VP Engineering, Human Signal

James was the first engineer at two acquired startups before joining AETHER. He brought a decade of embedded systems experience — from RTOS firmware to mobile SDKs — and rebuilt the Human Signal sensor fusion pipeline from scratch in 2023.

MH

Mira Holst

Head of Design

Mira designed systems at Linear (2020–2023) and Figma (2018–2020) before joining AETHER to build the design language for dense technical products. She believes that good design is mostly about removing things.

PN

Dr. Priya Nair

Head of Applied Science

Priya bridges research and product, running the team that turns research notes into production systems. Her background is in statistical signal processing and Bayesian inference. She published 14 papers before she was 30.

LE

Lars Eriksson

VP Engineering, Edge Lattice

Lars spent six years at Ericsson Research building mesh networking protocols before pivoting to the AI-adjacent layer. He wrote the original Edge Lattice consensus protocol over a single weekend in 2021, on a flight from Stockholm to Singapore.

AD

Amara Diallo

General Counsel

Amara joined from Linklaters where she specialized in technology contracts and cross-border data regulation. She drafted AETHER's privacy architecture and is the reason the company operates cleanly across 22 jurisdictions.