AI Alignment Engineer

about the company

Our client is a recognized pioneer in the governance and reliability of cognitive computing. They aim to define the global benchmarks for machine learning validation, ensuring that next-generation autonomous systems operate within secure, ethical, and transparent boundaries. Ensuring critical infrastructure needed to turn experimental AI into dependable enterprise solutions.

about the role

Looking for a specialist to engineer the methodologies required to rigorously validate and audit complex adaptive models within commercial environments. You will be responsible for defining the boundaries of how self-learning systems behave under pressure.

• Architect and implement assessment protocols to quantify the reliability and goal-alignment of autonomous software agents.

• Develop sophisticated statistical toolsets to measure predictive volatility and establish confidence intervals for model outputs.

• Build resilient diagnostic pipelines and testing environments for stochastic (non-deterministic) software architectures.

• Direct adversarial testing (red-teaming) and performance indexing using policy-gradient methods and human-in-the-loop refinements.

• Bridge the gap between environmental simulation and strategy deployment, focusing on the interplay of multi-component systems.

skills and experience

• Advanced degree in a quantitative discipline (e.g., Theoretical Physics, Applied Math, or Computational Theory).

• Expert-level command of the modern data science ecosystem, specifically (Python, NumPy, and PyTorch)

• Hands-on experience applying Bayesian inference techniques or non-parametric regression models (e.g., Gaussian Processes, Bayesian Neural Networks).

• Deep understanding of modern optimization strategies, including proximal policy methods and feedback-driven training.

• A track record of applying high-level engineering principles to the automation of complex validation suites

  
  To apply online please use the 'apply' function, alternatively you may contact Evangeline. (EA: 94C3609/ R24124002 )