Editorial summary

The company is seeking an AI Engineer responsible for experimentation and evaluation to understand business problems, define POC objectives, and establish evaluation metrics. The role involves designing experiments to test model configurations, prompts, or retrieval strategies, and analyzing Gen AI outputs for quality, accuracy, and alignment, while identifying common failure modes. It also covers data preparation and lightweight pipeline development, profiling datasets and building simple ingestion pipelines, with support for SMEs in defining ground truth benchmarks. The position offers a salary of 9200 and requires strong analytical skills and experience with Gen AI, data preparation, and pipeline development.

This summary is AI-generated and may contain inaccuracies. Please refer to the full job description below.

Job description

[This job https://singapore.job-q.com/jobs/detail/ai-engineer-38759 first appeared in Job-Q.com on 19 Oct 2026]

Responsibilities:

Experimentation & Evaluation

• Understand the business problem, POC objectives, and evaluation metrics.
• Design experiments to test different model configurations, prompts, or retrieval strategies.
• Analyse Gen AI outputs for quality, accuracy, and alignment with requirements; identify common failure modes (hallucination, bias, irrelevant answers, factual errors).
• Support SMEs in defining ground truth benchmarks for evaluation.

Data Preparation &Pipelines

• Profile and clean sample datasets for experimentation (lightweight data prep).
• Build and test simple pipelines for data ingestion, prompt construction and output evaluation.

Gen AI & Agentic Techniques

• Work with foundation models via AWS Bedrock, Google Vertex AI, or Azure AI Foundry depending on engagement cloud posture.
• Apply working knowledge of China-origin models (DeepSeek, Qwen, GLM) as increasingly relevant, cost-effective alternatives.
• Apply agentic orchestration frameworks such as AWS Strands, LangGraph, or equivalent, for designing and testing multi-step agent workflows.
• Apply prompt strategies, prompt engineering patterns, and RAG design (chunking, embeddings, retrieval evaluation); support ingesting/vectorising content to knowledge bases.
• Provide insights and recommendations to improve model performance in quick iterations, including fine-tuning approaches where applicable.

FDE &Development/Maintenance Coverage

• During FDE engagements: rapidly test candidate models, prompts, and retrieval strategies, giving the team fast, evidence-based go/no-go signals.
• During system development & maintenance engagements: support ongoing model/prompt tuning and monitoring as applications move toward production.

Collaboration

• Collaborate with developers on integrating models into the POC workflow, and work closely with PM, devs, and SMEs to refine data and prompts.
• Partner with the Data Scientist on evaluation methodology where classical statistical baselines are in play, and with the AI/LLM Specialist when an engagement moves toward production-grade evaluation.
• Document experiments briefly but clearly (hypothesis → result → conclusion).

Requirements:

• 3 years hands-on experience in AI/ML or Gen AI engineering
• Understanding of Gen AI concepts (tokenization, embeddings, RAG, prompting, evaluation).
• Familiar with at least one major cloud AI service (AWS Bedrock, Google Vertex AI, or Azure AI Foundry); working knowledge of others a plus.
• Working knowledge of the China AI model landscape (DeepSeek, Qwen, GLM) a strong plus.
• Familiarity with agentic orchestration frameworks (AWS Strands, LangGraph, or equivalent), for designing and testing multi-step agent workflows.
• Ability to do rapid experimentation rather than perfect models.
• Basic proficiency in Python and Gen AI tools (e.g., model SDKs, vector DBs).
• Analytical mindset: can quantify subjective output (accuracy, relevance, readability).
• Good data wrangling skills to prepare small datasets quickly.

Preferred Qualifications

• Generative AI Leader or Machine Learning Engineer certification, or equivalent.
• Exposure to LLMOps practices (model monitoring, versioning) for production transition.
• Familiarity with model fine-tuning techniques.
• Exposure to regulated government cloud environments.

Tech Stack(Illustrative)

• Languages: Python
• LLM Runtime: AWS Bedrock, Google Vertex AI, Azure AI Foundry; DeepSeek/Qwen/GLM (China stack)
• Agentic Frameworks: AWS Strands, LangGraph
• Data & Eval: Model SDKs, vector DBs, pandas/Jupyter-style tooling for experimentation

Interested candidates may send their CV to MAC (Reg No. R1221300) quoting the job title in the Subject line. We regret that only shortlisted candidates will be notified.

Scam prevention reminder: You should not make any pre-payment when applying for any job.

Illegal practices reminder: It is illegal for recruiter to collect payment (kickback) from the worker https://www.mom.gov.sg/-/media/mom/documents/publications/foreign-workers/what-are-kickbacks.pdf