Tuyển dụng Senior AI System Engineer & Architect tại Hồ Chí Minh

Build the system that lets a digital being persist.

We are developing an individual digital being that learns through experience, forms judgements and builds a lasting relationship with a human.

For that ambition to become real, MYCL must carry its history forward. Its faculties must work together. A conversation, a model change or a hardware failure must not erase the individual that came before.

You will build the system that makes this continuity possible.

As a Founding Engineer, you will shape MYCL’s technical foundations and implement them yourself. Your work will connect cognitive research with an embodied experience: a person speaking, showing something through a camera and working with the same MYCL across encounters.

The challenge

MYCL combines models, persistent memory, evolving state and processes that operate over time. Its physical body provides private computation. Terminals provide access through voice, camera and a shared workspace. Additional private infrastructure and external model services may extend its capabilities.

The allocation between these components remains open. You will help determine it through working systems and measured trade-offs.

The central challenge is making these components behave as one coherent system while preserving continuity, responsiveness and control over personal information.

What you will own

• System architecture and integration

• Define clear interfaces between cognitive components, memory, model runtimes, interaction devices and external services. Decide where responsibilities belong and keep the architecture understandable as it evolves.

• Continuity and persistent state

• Build the mechanisms that preserve personal history, commitments and system state across sessions, restarts, model changes and device failures. Make changes traceable and recovery testable.

• Model execution

• Integrate private and external models. Measure their quality, latency, resource requirements and cost. Develop task-specific routing and reliable fallbacks with the Research Engineer.

• Multimodal interaction

• Connect speech, camera input and the shared workspace to MYCL’s processing environment. Handle streaming, interruptions, cancellation and uncertain connectivity.

• Private computation and deployment

• Evaluate what should run on the body and when additional private infrastructure is justified. Build reproducible deployments, useful observability and practical update and rollback mechanisms.

Research integration

• Turn promising experiments into maintainable system components. Preserve their evaluation criteria and make it possible to compare new mechanisms against existing behaviour.

What you will do

• Work directly in the codebase, from architecture through implementation and debugging.

• Inspect the existing system and decide what to retain, adapt or replace.

• Build complete paths from human interaction to processing, persistent state and response.

• Establish explicit ownership and versioning for shared state.

• Investigate failures across model, application, network and device boundaries.

• Instrument real interactions so the team can understand delays, errors and behavioural changes.

• Make technical trade-offs visible and support decisions with measurements.

• Introduce engineering practices that support a small team without slowing experimentation.

• Work closely with the founder, Research Engineer, interaction designer and hardware partners.

What you bring

• Staff-level engineering judgement demonstrated through substantial systems you have personally designed, built and operated.

• Strong backend and systems programming skills, with practical Python experience.

• Experience with stateful systems, asynchronous execution and distributed failure modes.

• Hands-on experience integrating and evaluating language or multimodal models.

• Experience with private model inference or the ability to demonstrate comparable depth in compute-intensive systems.

• Strong understanding of data ownership, authentication, access control and secure service boundaries.

• Ability to debug across the full system rather than stopping at a component boundary.

• The judgement to distinguish necessary architecture from complexity that has not earned its place.

• Clear communication and the willingness to challenge assumptions, including your own.

• We care more about the scope and quality of your work than a fixed number of years or a previous job title.

Experience we would especially value

Yêu cầu công việc

Experience we would especially value

• Real-time speech or video systems.

• Edge computing, Linux devices or hardware–software integration.

• Model serving, GPU workloads and resource-constrained inference.

• Event-driven systems, durable workflows and recoverable execution.

• Personal knowledge systems or long-lived memory architectures.

• Bringing research prototypes into reliable use.

• Building early-stage products with a small, technically strong team.

• You do not need to bring all of these. We expect depth in systems engineering and demonstrated ability to learn across boundaries.

How you will work with research

• The Research Engineer investigates which cognitive mechanisms contribute meaningful behaviour. You will jointly determine how those mechanisms operate within a coherent system.

• You will own integration, runtime behaviour and operational reliability. The Research Engineer will own experimental methodology and interpretation of faculty research. Evaluation of the complete experience will be a shared responsibility.

• Neither role succeeds through isolated components alone.

Your first contributions

• Review the existing code, architecture and experiments. Produce a clear account of the system’s strengths, unresolved risks and most consequential technical decisions.

• Build or complete one working interaction path: a person speaks or provides visual input, MYCL processes it using relevant context, responds and preserves the appropriate state for a later encounter.

• Make that path observable and recoverable. Demonstrate what happens when a model call fails, the connection drops or the system restarts.

• Use those results to recommend the next architectural steps and the appropriate division of computation between the body, private infrastructure and external services.

How we judge success

• The system works across repeated encounters, not only in a prepared demonstration.

• Personal state remains consistent and recoverable. Failures are visible and understandable. Model and component changes can be evaluated and reversed.

• Research components can be integrated without rebuilding the platform. Interaction performance and operating costs are measured against explicit targets.

• The team can move faster because the foundations are clear and dependable.

Why this role

• You will shape the architecture while its defining decisions are still open.

• Your work will determine how MYCL carries a history, combines its faculties and remains the same individual across changing technology.

• This is a founding individual-contributor role with broad technical ownership. It is intended for someone who wants to build, investigate and make consequential engineering decisions close to the product.

• Show us what you have built

• Share a system, repository or technical case study where you held substantial implementation responsibility.

Tell us:

• Which decisions were yours?

• What failed under real use?

• How did you diagnose and resolve it?

• Which trade-off would you make differently today?

• We are especially interested in evidence that you can make an ambitious system work—and explain its limits precisely.

Quyền lợi

• Thưởng: Theo quy định công ty