Applied AI Engineer

Onebyzero; Singapore; On-site; SGD 180,000 - 260,000; Full time

Job summary

Onebyzero in Singapore is seeking a Deep Learning Architect with 4+ years to design production-grade GenAI systems, focusing on end-to-end LLM architectures, RAG pipelines, and multi-agent setups with strong production readiness.

You will translate client requirements into model adaptation strategies, review designs, optimise latency and cost, and collaborate with engineering and data teams to deliver high-impact GenAI solutions.

Qualifications

• 3–6 years of ML engineering/LLM experience with production focus.
• Hands-on experience with CPT, SFT, LoRA or QLoRA on LLMs.
• Strong Python programming for production-grade code.
• Experience deploying models to cloud with cost and latency awareness.
• Ability to translate client requirements into model adaptation strategies.

Responsibilities

• Design and contribute to end-to-end LLM system architecture for real-world enterprise use cases.
• Pre-train, fine-tune LLMs and domain models using CPT, SFT, LoRA, QLoRA.
• Build evaluation pipelines to benchmark performance, accuracy, and cost.
• Optimise latency, throughput, token efficiency, and inference cost in production.
• Integrate fine-tuned models into multi-agent pipelines with orchestration teams.
• Implement prompt versioning, rollback strategies, and model monitoring.
• Translate business requirements into model adaptation strategies with clear success criteria.
• Contribute to internal knowledge sharing on fine-tuning practices and tooling.
• Define enterprise patterns for GenAI systems: identity, governance, and data boundaries.

Tools

Bedrock

OpenSearch

Lambda

ECS/EKS

Pinecone

Weaviate

Milvus

Docker

Kubernetes

Job description

About the Role

We are seeking a Deep Learning Architect with 4+ years of experience to help design and build production-grade GenAI systems. In this role, you will contribute architecture coverage across the team—reviewing system designs, identifying gaps, and guiding technical decisions at the solution level. You will work on end-to-end LLM system design, Retrieval-Augmented Generation (RAG) pipelines, and multi-agent architectures, with a strong focus on production readiness. Strong coding depth is non-negotiable.

Responsibilities

• Design and contribute to end-to-end LLM system architecture for real-world enterprise use cases (from requirements to production).
• Pre-train, fine-tune LLMs and domain-specific models using techniques such as CPT, SFT, LoRA, and QLoRA for client-specific use cases.
• Design and run model evaluation pipelines to benchmark performance, accuracy, and cost across different fine-tuning approaches.
• Optimise models for latency, throughput, token efficiency, and inference cost in production environments.
• Work alongside agent orchestration and architecture teams to integrate fine-tuned models into multi-agent pipelines.
• Implement prompt versioning, rollback strategies, and model monitoring to ensure reliability post-deployment.
• Translate business requirements from client engagements into model adaptation strategies with clear success criteria.
• Contribute to internal knowledge sharing on fine-tuning best practices, tooling, and emerging techniques.
• Define enterprise integration patterns for GenAI systems (identity/access controls, auditability, data boundaries, governance, and compliance alignment).
• Improve production reliability: latency/throughput optimization, token efficiency, cost control, and robust failure handling.
• Collaborate with cross-functional stakeholders (engineering, data, product, client teams) to deliver high-impact solutions on tight timelines.
• Contribute hands-on code, perform code reviews, and raise the engineering bar through strong software fundamentals.

Qualifications

• 3–6 years of experience in ML engineering, LLMs, or model development roles.
• Hands-on experience with Continual Pre-training (CPT), Supervised Fine-tuning (SFT), LoRA, or QLoRA on LLMs.
• Strong Python programming skills—ability to write clean, testable, production-ready code.
• Experience running model evaluation and benchmarking pipelines in a structured way.
• Solid understanding of transformer architectures and how fine-tuning affects model behaviour.
• Experience deploying fine-tuned and pre-trained models to cloud environments with attention to cost and latency.
• Strong problem-solving skills with the ability to work independently on client-facing projects.
• Solid software engineering fundamentals: APIs, data structures, testing, debugging, and performance optimization.
• Ability to review designs, communicate trade-offs clearly, and collaborate effectively in a fast-paced environment.

Required Skills

• Experience with AWS-native GenAI building blocks (e.g., Bedrock, OpenSearch, Lambda, ECS/EKS) and secure enterprise deployments.
• Experience with vector databases/search engines (OpenSearch, Pinecone, Weaviate, Milvus, FAISS) and retrieval optimization.
• Experience with containerization and orchestration (Docker, Kubernetes).
• Experience building evaluation/observability pipelines for LLM systems and implementing safety/guardrail patterns.
• Consulting or client-facing delivery experience.