Building Practical On-Device AI for Engineering Teams
We started Skerry Labs because product teams kept running into the same wall: compelling AI demos that didn't translate well onto actual hardware.
← Back to HomeHow Skerry Labs Came Together
Skerry Labs was founded in Shah Alam in 2021 by a small group of engineers who had spent years working on embedded systems and machine learning infrastructure. The team noticed a recurring pattern: product companies in Malaysia were building hardware that was capable of running local inference, but they were routing analysis through cloud endpoints anyway — not because it was the better approach, but because setting up on-device AI was genuinely difficult to get right.
We set out to close that gap. Not by offering a platform or a subscription product, but by working directly with engineering leads and product teams to configure, document, and hand over on-device deployments they could maintain themselves.
The name comes from a skerry — a small, exposed rocky outcrop in the sea. It's a practical kind of landmark: precise, useful, and easy to navigate from. That's what we want our work to be for the teams we engage with.
What We're Here to Do
Our aim is to make on-device AI a dependable, well-understood part of the products our clients build — not a black box that only specialists can manage. Every engagement ends with the client team holding clear documentation and a maintainable deployment.
Scoped Engagements
We match the service to where you actually are — exploring, setting up a single device, or scaling to a fleet. No over-engineering.
Documentation First
Every deployment comes with written handover materials — update procedures, review schedules, configuration notes — so you're not dependent on us after the engagement ends.
Team Knowledge Transfer
We run working sessions with your engineers so the on-device setup is understood by the people who will live with it day-to-day.
Who Works on Your Projects
A compact team with backgrounds in embedded systems, edge computing, and AI model optimisation.
Zulaikha Faris
Founder & Edge Systems Lead
Eight years in embedded Linux and hardware-adjacent AI before starting Skerry Labs. Leads technical scoping and on-device configuration work.
Rajan Nair
Model Optimisation Engineer
Focuses on quantisation and inference pipeline tuning, making standard models fit comfortably within constrained device environments.
Yi Lin Tan
Deployment & Documentation Specialist
Handles fleet coordination, update routine design, and produces the maintenance handbooks that come with every On-Device Setup and Programme engagement.
Our Working Standards
These principles shape how we approach every engagement, regardless of service tier.
Honest Scoping
We tell you clearly if a use case isn't a good fit for on-device processing — before you spend money on setup work. The feasibility review exists specifically for this.
Data Handling Care
We don't retain client device data or model artefacts after an engagement closes. Configurations and handover materials are delivered to you and not stored on our side.
Version-Controlled Deliverables
All configuration files, update scripts, and documentation are delivered in version-controlled form so your team can track changes and roll back if needed.
Knowledge Transfer
Every setup service includes a session where we walk your engineers through what was done, why, and how to maintain it. We want your team confident, not dependent.
Defined Update Routines
On-device deployments need a clear update path. We design and document the update routine as part of the engagement so model refreshes don't require outside help.
Direct Communication
You'll work directly with the engineers doing the work — no account management layer. Questions get answered by the people with the relevant context.
Edge AI Deployment in Malaysia — What Skerry Labs Focuses On
On-device AI deployment is a distinct discipline from standard machine learning development. The constraints are real: memory budgets measured in megabytes, compute resources a fraction of what a server provides, update cycles that need to work reliably over cellular or Wi-Fi rather than a data centre uplink. Skerry Labs has built its practice specifically around these realities, working with hardware ranging from microcontroller-class devices to embedded single-board computers.
Malaysian engineering teams building IoT products, industrial sensors, or smart devices face a common inflection point: the moment when cloud-based AI stops being the right default. Connectivity is not always dependable. Data regulations around personal and operational data are developing. Cloud inference costs accumulate as device fleets grow. On-device processing addresses all three of these in a straightforward way — but only if the deployment is set up correctly from the start.
Skerry Labs works with teams at the feasibility stage — before commitments are made — through to fleet-wide deployments with structured review cycles. The focus throughout is on setups that the client team can understand, maintain, and update without external assistance.
Want to Learn More?
Get in touch to discuss your device setup, use case, or any questions about how on-device AI might fit your product.
Contact Skerry Labs