AI Model Trainer

CHRONOAI PTE. LTD.; Singapore; Presencial; SGD 60.000 - 90.000; Jornada completa

Descripción de la vacante

ChronoAI is seeking an AI Model Trainer in Singapore to advance model capabilities through better data, training methods and evaluation. You will turn complex knowledge into structured training data and design experiments from dataset design to evaluation.

You will collaborate with AI/ML engineers and leadership, own hypotheses, and drive hands-on training, fine-tuning and evaluation of language and reasoning models.

Formación

• Degree in CS/AI/DS/Math/Stats/Engineering/Physics or related.
• Hands-on ML experience with training, fine-tuning, data preparation or evaluation.
• Proficient in Python and ML frameworks.
• Ability to design experiments and reason about failures.
• Experience with language/model training is a plus.

Responsabilidades

• Develop, curate and validate high-quality datasets for model training.
• Translate domain knowledge into machine-trainable examples.
• Run training, fine-tuning and iterative optimisation experiments.
• Design evaluation criteria and benchmarks.
• Analyse model outputs for accuracy and reasoning quality.
• Identify failure patterns and hypothesise causes.
• Experiment with prompting, synthetic data and model selection.
• Ensure data quality and avoid error amplification at scale.
• Document datasets, methods and results for reproducibility.
• Use AI tools to accelerate repetitive work with human oversight.

Conocimientos

Machine learning

Python

Data preparation

Evaluation

Analytical reasoning

Experiment design

Reproducibility

PyTorch

Hugging Face

LoRA/QLoRA

Educación

Bachelor's degree in CS/AI/DS/Math/Stats/Engineering/Physics

Statistics/Math background

Herramientas

PyTorch

Hugging Face Transformers

LoRA/QLoRA

PEFT

Descripción del empleo

About the role

We are looking for an AI Model Trainer to improve model capabilities through better data, training methods and evaluation. This is a hands-on, experimental role. You will turn complex knowledge and reasoning processes into structured training examples, investigate where models fail, and design experiments to determine what actually improves performance. You will work with domain experts, AI/ML engineers and leadership, but you should be comfortable forming your own hypotheses and taking ownership of the work from dataset design through to evaluation.

Key responsibilities

• Develop, curate and validate high-quality datasets for model training and fine-tuning.
• Translate abstract concepts, domain expertise and multi-step reasoning into clear, machine-trainable examples.
• Run training, fine-tuning and iterative optimisation experiments.
• Design evaluation criteria, test cases and benchmarks that measure the capabilities we actually need.
• Analyse model outputs for accuracy, reasoning quality, consistency and completeness.
• Identify hallucinations, reasoning errors, knowledge gaps and recurring failure patterns; investigate their likely causes.
• Experiment with prompting, synthetic data, training approaches and model selection, using results to decide what to try next.
• Put quality controls around data generation and evaluation so errors are not amplified at scale.
• Document datasets, methods, experiment results and decisions clearly enough for the team to reproduce and build on them.
• Use AI tools and automation to accelerate repetitive work while maintaining rigorous human judgment over data quality and conclusions.

About you

• A degree in Computer Science, AI, Data Science, Mathematics, Statistics, Engineering, Physics or a related discipline.
• A strong understanding of machine learning and modern language models, with hands‑on experience or solid practical knowledge of training, fine‑tuning, data preparation or evaluation.
• Python proficiency and familiarity with common AI/ML frameworks and workflows.
• Strong analytical and logical reasoning skills, including the ability to structure complex or abstract knowledge.
• The ability to examine individual failures, not just aggregate metrics, and turn observations into testable hypotheses.
• Care with experimental design, data quality, documentation and reproducibility.
• Curiosity about new AI methods, comfort with ambiguity and the self‑direction to make progress in a fast‑moving environment.
• Experience training, fine‑tuning or evaluating language and reasoning models (nice to have).
• Familiarity with PyTorch, Hugging Face Transformers, LoRA/QLoRA, PEFT or similar tools (nice to have).
• Exposure to SFT, reinforcement learning, synthetic data generation or reasoning‑model training (nice to have).

About us

ChronoAI is a Singapore‑headquartered AI technology company building infrastructure for the next generation of AI agents. Our mission is to make persistent, always‑on AI agents as easy to deploy as cloud services. We are building an Agent‑as‑a‑Service ecosystem covering connectivity, runtime and skills. Our products include NyxID, which connects AI agents to services and applications; Aevatar, infrastructure for persistent cloud‑based agents; Ornn, a plug‑and‑play skill marketplace for AI agents; and Automath, an AI engine for mathematical discovery and formal verification. At ChronoAI, we believe AI should handle more execution while humans focus on judgment, creativity and decision‑making. We value people who are curious, move fast, take ownership and constantly look for smarter ways to solve problems.