Founding Research Scientist, Forecasting Foundation Models

About Torchcast AI

Torchcast AI is a Singapore-based company that builds AI systems to forecast real-world outcomes. Our platform combines forecasting-trained models, evidence retrieval, quantitative methods and probabilistic reasoning to help organisations make better decisions under uncertainty. We are developing Forecasting Foundation Models (FFM): models purpose-built to learn from diverse evidence, reason under uncertainty and produce accurate probability forecasts.

The role

As a Founding Research Scientist in our AI Labs team, you will help build the FFM research programme. You will work directly with the founders and engineering team and own research projects from formulation through experimentation, evaluation and deployment. This is a hands-on role with significant influence over research direction, technical standards and future hiring.

What you will do

• Research and develop new machine-learning methods for probabilistic forecasting.
• Build models that learn from text, time series, structured data, market signals and outputs from other models.
• Develop training and evaluation datasets from timestamped evidence, historical forecasts and resolved outcomes.
• Design rigorous experiments against strong language-model, quantitative, time-series and ensemble baselines.
• Evaluate models for accuracy, calibration, robustness and generalisation.
• Investigate forecasting-specific training objectives, architectures and representation-learning methods.
• Work with engineering to scale promising research into production systems.
• Contribute to technical reports, publications, benchmarks and open-source releases where appropriate.

Requirements

• Strong track record in machine-learning research or in building technically novel ML systems.
• Deep expertise in at least one relevant area, such as probabilistic machine learning, representation learning, multimodal learning, time-series modelling, foundation models, uncertainty estimation or structured prediction.
• Strong programming skills and practical experience with PyTorch, JAX or a comparable framework.
• Solid understanding of experimental design, model evaluation, temporal validation and data leakage.
• Ability to take an open-ended research problem from formulation to rigorous evaluation, and to work independently in an early-stage environment.
• A PhD in machine learning, computer science, statistics, applied mathematics or a related field is preferred but not required.

Nice to have

Experience with probabilistic forecasting, calibration or proper scoring rules; multi-source or multimodal learning; training or adapting large-scale models; reinforcement learning; financial, economic or geopolitical forecasting; publications at leading ML or statistics venues; moving research into production.

How to apply

Apply through MyCareersFuture or email [email protected] with your CV, links to relevant papers or code, and a response of no more than 300 words to: "What is one important research question in AI forecasting that you would investigate, and how would you begin testing it?"

Torchcast AI is an equal opportunity employer. Selection is based on skills, experience and ability to do the job.