Senior AI Architect & Systems Engineer

ARRK Engineering; Petaling Jaya; On-site; MYR 150,000 - 210,000; Full time

Job summary

ARRK Engineering in Malaysia seeks an experienced software engineer to join our AI initiative, designing, coding, and deploying a suite of intelligent AI tools and production-ready components. You will work across multi-country teams to optimize context windows, token usage, and retrieval architectures, with emphasis on security, compliance, and high-quality delivery.

The role requires hands-on Python expertise, experience with vector databases and graph representations, and a track record of

Qualifications

• 5+ years in software engineering with substantial AI/ML product delivery.
• Strong conceptual grasp of how transformers process data - tokens, self-attention, embeddings, and context window limitations.
• Hands on experience with vector databases and hybrid retrieval (dense + sparse, reranking, fusion); familiarity with multimodal/visual retrieval is a plus.
• Experience with graph based representations of code or documents (property graphs, AST parsers, libraries such as Tree sitter / NetworkX) to programmatically read, prune, and structure data for LLM ingestion.
• Familiarity with both self hosted model serving (e.g. vLLM, Ollama) and cloud-managed AI services - and a clear sense of when each is appropriate, especially under data sovereignty constraints.
• Excellent communication skills in English; German is a plus given our multi-country setup.
• Awareness of security and compliance requirements in regulated environments (e.g. TISAX, EU AI Act, NIS2/DORA) and how they shape AI architecture.
• Experience with fine tuning / adapter approaches (e.g. LoRA) and structured model evaluation is a plus.

Responsibilities

• Act as a primary technical counterpart and architecture partner - designing, coding, and deploying a suite of custom, intelligent AI tools, and critically reviewing design decisions.
• Enhance our system's ability to manage context windows, optimize token usage, and reduce attention noise when connecting LLMs to large codebase repositories.
• Design and extend graph based code and document representations (e.g. property graphs, AST derived structures) and hybrid retrieval pipelines (dense + sparse + reranking/fusion) over vector databases.
• Translate complex algorithmic ideas - syntax parsing, semantic chunking, custom routing - into reliable, production ready software components.
• Build and rigorously evaluate prototypes for future AI initiatives, including open source small language models (SLMs), custom embeddings, and multi agent / cross validation patterns, backed by measurable retrieval and quality metrics.
• Architect multi tenant data, identity, and authorization layers that meet enterprise security and compliance requirements.
• Ensure our AI applications integrate securely into existing codebase ecosystems and developer toolchains, with both self hosted and cloud managed model serving in mind

Skills

AI/ML product delivery

Backend systems

Python

Transformers concepts

Vector databases

Graph representations

Cloud platforms

Security & compliance

English communication

Tools

Tree sitter

NetworkX

vLLM

Ollama

Job description

ARRK Engineering in Malaysia seeks an experienced software engineer to join our AI initiative, designing, coding, and deploying a suite of intelligent AI tools and production-ready components. You will work across multi-country teams to optimize context windows, token usage, and retrieval architectures, with emphasis on security, compliance, and high-quality delivery.

The role requires hands-on Python expertise, experience with vector databases and graph representations, and a track record of