Lead AI Scientist: Autonomous Agents & Planning (Remote)

Humana; United States; On-site; USD 142,000 - 196,000; Full time

Benefits offered by this job

Health benefits

Bonus incentive plan

Remote work (US)

Job summary

Humana is seeking an expert in agent design to develop and evaluate autonomous agents capable of setting, adapting, and executing goals in real-world environments. You will contribute to scalable AI modules, safety, and evaluation pipelines within the Enterprise AI organization.

The role emphasizes planning, control, and collaboration across teams, with a focus on interpretable, robust, and production-ready AI systems.

Qualifications

• Master's Degree and 4+ years of research/ML engineering or applied research scientist experience.

Responsibilities

• Architect goal setting and decomposition for autonomous agents in uncertain environments.
• Design and implement dynamic planning systems (hierarchical planning, curriculum learning, scratchpad loops).
• Collaborate on memory, tool-use, and feedback-loop designs for multi-step agent behavior.
• Develop evaluation frameworks to align with human intent and long-horizon objectives.
• Prototype agents interfacing with APIs, MCP servers, search engines, databases, or actuators.
• Explore mechanisms to detect goal misalignment and unsafe emergent strategies.
• Work with teams to integrate goal alignment with safety and reliability mechanisms.

Skills

SQL

Python

Data analysis

ML frameworks

Transformers

Symbolic planning

ETL

Large-scale ML

Education

Master's degree or higher

Ph.D. in CS/DS/ML or related field

Tools

PyTorch

JAX

LangChain

LangGraph

AutoGen

Job description

Humana is seeking an expert in agent design to develop and evaluate autonomous agents capable of setting, adapting, and executing goals in real-world environments. You will contribute to scalable AI modules, safety, and evaluation pipelines within the Enterprise AI organization.

The role emphasizes planning, control, and collaboration across teams, with a focus on interpretable, robust, and production-ready AI systems.