Senior ML Architect — LLM Reasoning & Tools (Remote)

Alignerr Corp.; Sydney; A distancia; AUD 207.000 - 344.000; A tiempo parcial

Descripción de la vacante

Alignerr is seeking a Senior Machine Learning Expert to author high-fidelity reasoning traces that guide how an LLM should plan, reason, and act step by step. This remote contract role emphasizes independent, asynchronous work on advanced AI problems.

You will design structured traces, review others' work for quality, and develop data strategies to improve model reasoning and reliability. Strong background in ML evaluation/training and RLHF workflows is preferred.

Formación

• Experienced ML engineer or researcher with a strong background in model reasoning, evaluation, or training.
• Skilled at breaking down complex technical problems into clear, structured steps.
• Familiar with advanced LLM evaluation and training methodologies.
• Detail‑oriented with a systematic approach and comfort working asynchronously in a remote environment.

Responsabilidades

• Author high-fidelity reasoning traces that guide LLM planning, reasoning, and action step by step.
• Design structured traces documenting tool use, decision making, and problem decomposition.
• Review traces authored by others to ensure quality and consistency.
• Develop data strategies to help LLMs navigate real‑world scenarios with accuracy and dependability.
• Apply senior ML architectural insights to improve model reasoning across open‑ended tasks.

Conocimientos

Machine Learning

LLM evaluation

Problem decomposition

Independent work

Herramientas

Kaggle

Data labeling tools

Descripción del empleo

Alignerr is seeking a Senior Machine Learning Expert to author high-fidelity reasoning traces that guide how an LLM should plan, reason, and act step by step. This remote contract role emphasizes independent, asynchronous work on advanced AI problems.

You will design structured traces, review others' work for quality, and develop data strategies to improve model reasoning and reliability. Strong background in ML evaluation/training and RLHF workflows is preferred.