居家办公 Machine Learning Researcher 工作, 薪水 up to HKD 23,000, M80 Limited Hong Kong 公司招聘中 - Ricebowl

About m80

m80 is an AI-native healthspan company in Hong Kong. We build agents that own a health domain end-to-end and act on it. The agent has to hold a coherent model of one person over months, plan across horizons that disagree, and change its own plan on evidence without losing safety or validity.

The Role

You run the model-side investigation of a paired research plan, across three problems:

• Per-user context representation — schemas, signal extraction and weighting, consolidation and forgetting, retrieval.
• Multi-horizon agentic planning — goal decomposition, cross-horizon plan representation, replanning triggers, consistency maintenance.
• Closed-loop adaptation — outcome measurement, autonomous protocols under adherence variability and confounding.

You own them end-to-end: design, implementation, evaluation, characterisation, write-up. That includes building the evaluation corpora and harnesses, running comparative evaluations and ablations against a seven-axis framework, maintaining the interface with the paired systems investigation, and delivering on a regular research cadence ending in a technical white paper.

You'll need

• MSc or PhD in CS, ML, statistics, machine learning or a close STEM field.
• Strong applied ML/AI engineering with Python, modern LLM/agent tooling, and the ability to take a method from idea to measured implementation.
• Experimental discipline: you design the evaluation before you build the thing, and can defend a result against an ablation.
• Previous research experience will be a plus

Advantages: agentic systems, LLM planning or memory architectures; causal inference under confounding; wearable and physiological time-series data; relevant publications.

What you get

Longitudinal experimental practice, single-subject methodology, sustained characterisation of a full closed loop, and intended open-source release at a company small enough that your findings go straight into what gets built.

Apply

CV, a short note on which of the three problems you'd most want to own and why, to *************

薪酬： 每月 $23,000.00

Work Location: Hybrid remote in Hong Kong