Work

What we build.

We build frontier AI models and security systems for the communities that need them most.

LAMBA. In training · 2026

A frontier-tier general-purpose reasoning model that also speaks Bidayuh Bau — the world’s first. Currently in training.

“A line of Bidayuh Bau will live here.”


View LAMBA →

Also from the lab

AEGIS.Live

Physics-based adversarial network traffic detection.

View AEGIS →

HBEE.Research preview

Multi-agent sociotechnical insider-threat simulation.

Read the preprint →

Research

The record.

The research is the product. Every claim here is backed by a paper, a dataset, or a public benchmark — nothing unverifiable.

arXiv:2604.02149 AEGIS — thermodynamic state-space models for zero-day evasion detectioncs.CR · cs.LG · Published Read →
arXiv:2605.07472 HBEE — multi-agent sociotechnical simulation for insider-threat modelingcs.CR · Published Read →
Dataset AEGIS Adversarial Corpus908,037 sequences · gated HuggingFace →

Vision

A community that owns its technology and its future.

Mission

Advance quality of life through AI and deep tech — built for our people, by our people.

Founder

The builder.

Vickson Ferrel

Founder · Vixero Technology Enterprise

Kuching, Sarawak · UNIMAS FCSIT · Founded March 2026

Two sole-authored papers across two fields — cybersecurity and sociotechnical simulation — published without institutional support. Built independently — by design. A native Bidayuh Bau speaker, he is the sole validator for the world’s first frontier reasoning model in the language, with a living community of 200,000 speakers behind it.

LAMBA is being built because the AI transition is happening without Sarawak. Every frontier model in the world was trained without a single sentence of Bidayuh Bau. That’s not an oversight — it’s a structural gap that only someone from inside the community can close. The goal is not preservation for its own sake. It’s proof that a small community from East Malaysia can build at frontier level — and in doing so, create a replicable model for every other low-resource language in SEA that has been written out of the AI future. Sarawak has never appeared in an LLM benchmark table. LAMBA will change that.

ORCID 0009-0009-0155-0913