Apply: email your CV (PDF) plus a note of no more than ten lines to[email protected], subject line “Application: AI Platform Engineer”. In the note, say which models you have deployed, the hardware (GPU model or equivalent), and what you owned. Public repos or technical writing links welcome. Send longer attachments by email; the contact form is a fallback only — do not assume it can carry a CV. Do not attach an ID number, bank account or other sensitive personal data unrelated to screening on first apply.

About us

Infraplus Solutions Limited does enterprise network equipment procurement and delivery in Hong Kong and the Greater Bay Area: new, refurbished and used Cisco／Juniper／Huawei, compatible optics, sourcing and inspection, installation and setup, network engineering outsourcing, ITAD clearance, and certification labs. We are not an authorised reseller of any brand, and not a training academy. Existing GPU／data-centre pages on this site are about networking, optics and room procurement — not selling “LLM managed operations” as a product.

This role is engineering capability we are hiring for: AI Platform and LLMOps — deploy, maintain and wire models on our own AI infrastructure to apps and agents. If a later customer project needs similar work, that scope will be written into that engagement; do not read this page as a service you can order today.

Role

Chinese title: AI 平台工程師
English title: AI Platform Engineer

Own AI Platform and LLMOps on our private infrastructure: install, deploy, integrate and troubleshoot LLMs; capacity planning; RAG and guardrails.

What you'll do

  • AI Platform and LLMOps: install, deploy, integrate and troubleshoot open-source large language models and inference services in local, virtualised or container environments (for example Linux containers exposing an API). Open-weight models may include Llama, Qwen, DeepSeek, Mistral, Gemma or peers — whatever we actually run; you need not have done every family.
  • Triage outages, VRAM pressure, drivers, storage and network issues, and leave runbooks a colleague can take over.
  • Help with capacity planning (GPU, memory, storage and network); you may work with network／optics colleagues on selection, but this role is not colo power or cabinet contracting.
  • Enable RAG and private knowledge bases, plus basic guardrails (prompt injection and data-leak risk); tool calling and agents as needed.
  • Fine-tune or lightly adapt models when needed (LoRA／QLoRA／PEFT), with repeatable evaluation of quality, latency, throughput and resource use — not research-from-scratch training.
  • Write deployment notes, runbooks and incident records in English; Chinese documents as the team needs.

Must-haves

  • English required — explain technical issues, write runbooks, and work with vendors／docs in English without friction.
  • Linux; Docker／Kubernetes; open-source frameworks; basic GPU literacy (VRAM, batching, quantisation, driver／framework versions).
  • Real experience deploying and keeping open-source LLMs running on servers or in container／virtualised environments — not public-API-only work.
  • Production Python and wiring models to apps or internal flows via APIs.
  • Basic privacy awareness: on-prem often exists to keep data in a controlled environment.
  • No hard degree gate; show the above via portfolio, repos or verifiable projects. About 4 years of IT, including about 2 years in AI／generative AI; negotiable if you can show solid on-prem deployments.

Nice to have

  • Cantonese or Mandarin a plus (Hong Kong colleagues and Greater Bay Area counterparts).
  • RAG, vector stores, agent orchestration or tool calling.
  • Domain adaptation with LoRA／PEFT, or offline／online eval harnesses.
  • Local inference stacks such as vLLM, TGI or peers, and the open-model ecosystem.
  • GPU servers, high-speed networking or room hands-on experience.
  • Security and basic guardrails. Certifications are not a hiring requirement.

What this role is not

  • Not a bank Head of AI, and not a pure cloud data-science lead role.
  • Not a sales seat selling externally the service of deploying or operating LLMs for customers.
  • Not a training lecturer role.

Location and application

Based in Hong Kong; five-day work week (Monday–Friday). Exact location is confirmed at hire; the site does not publish an address, phone number or WhatsApp. Email your CV (PDF) and a short note to[email protected](subject: Application: AI Platform Engineer). We reply to the same address; office／support hours Mon–Fri 09:00–18:00 Hong Kong time, without promising same-day replies to every application.

Compensation and benefits

  • Base salary: HKD 45,000–65,000 per month, positioned within that range by experience; the final point is stated in the offer letter.
  • 13th month: the contract provides for a 13th-month salary; payout timing follows the contract.
  • MPF: statutory scheme; employer and employee each contribute 5% of relevant income within MPFA caps.
  • Medical plan: company medical cover; scope is confirmed at hire.
  • Discretionary bonus depends on company and individual performance and is not guaranteed year to year; any extra-month bonus not committed in writing is not treated as fixed.

Equal opportunity and applicant privacy

We consider qualified applicants on ability, experience and the needs of this role, without less favourable treatment on grounds of sex, age, race, nationality, religion, disability, family status or sexual orientation. Application data received by email or form is used only for this hire and handled under Hong Kong’s Personal Data (Privacy) Ordinance. After review we delete or anonymise under the retention policy in force; we do not use applicant data for marketing. Detail in the privacy policy.

Related pages