About Time Seventeen
We give risk a price. We create state prices for tail events.
The world contains a large class of risks that determine where capital flows yet have no price today. Global supply chains are being reconfigured, while the geopolitical discount rates that inform consequential decisions remain difficult to quantify. Time Seventeen is building tradable synthetic indices and term structures for these risks.
Time Seventeen is infrastructure for geopolitical risk and catastrophe risk. We give risk a price. We create state prices for tail events. Our ambition is to turn fragmented judgments about individual events into continuous risk prices, then bring those prices into institutional valuation, trading, and risk-transfer workflows.
Our work connects event definitions, probabilistic forecasting, market data, index methodology, and execution infrastructure. We explore how prediction markets, actuarial models, insurance and reinsurance structures, and blockchain-based DeFi can contribute to a common language for uncertain outcomes. The challenge is to make risk legible across different horizons, exposures, and market participants, while keeping assumptions and limitations explicit.
Building this infrastructure requires more than a compelling forecast. Data must be traceable, events must have defensible resolution rules, methodologies must survive scrutiny, and products must account for liquidity, basis risk, operational resilience, and applicable legal constraints. We want institutions to be able to understand a risk price, challenge its construction, and assess how it relates to the exposure they actually hold.
We are building a focused team across research, engineering, markets, and risk. We value people who can move between deep specialist work and the practical decisions needed to build a company: identifying the important question, testing an approach, communicating the evidence, and carrying the result into a usable system.
About the role
Investigate AI methods for geopolitical and catastrophe forecasting, with an emphasis on calibration, evidence quality, and reproducible evaluation.
You will study how AI can help interpret evolving events and produce useful inputs for probabilistic risk research. Projects may involve information extraction, retrieval, event classification, forecasting assistance, or scenario generation. The goal is to test whether a method improves a defined task, rather than merely producing plausible text.
Working with researchers and engineers, you will build experiments with explicit baselines and evaluation criteria. You will examine uncertainty, source reliability, temporal leakage, and failure cases, then communicate which findings can support a product decision and which need more evidence.
What you'll do
- Review relevant research and formulate a focused, testable hypothesis for an event-risk task.
- Prepare versioned datasets with documented provenance, labeling rules, and time-based evaluation splits.
- Prototype language-model, retrieval, or statistical approaches using Python and appropriate tools.
- Evaluate calibration, robustness, error patterns, and performance against meaningful baselines.
- Analyze failure cases such as unsupported claims, stale evidence, inconsistent event definitions, and data leakage.
- Present reproducible experiments and concise research notes that distinguish results from speculation.
What we're looking for
- Student or early-career researcher in computer science, statistics, machine learning, economics, or a related discipline.
- Practical Python skills and familiarity with core machine-learning and probability concepts.
- Experience with an AI project, research paper, competition, or substantial analytical coursework.
- Interest in evaluation and willingness to investigate results that challenge the initial hypothesis.
- Clear written communication and ability to work independently on a bounded research question.
What success looks like
- Deliver a reproducible experiment with a documented dataset, baseline, and evaluation method.
- Identify where the approach adds useful information and where it fails.
- Share research artifacts that another team member can inspect and extend.
Location & working arrangement
This role is associated with San Francisco and supports remote work. Please share your location, time zone, availability, and ability to coordinate with colleagues in San Francisco and Hong Kong.
How to apply
Share your CV and a research project, code repository, analytical report, or other example of work you can discuss. Explain your contribution and what you learned. Please include your preferred start date and internship availability.
How we work
Time Seventeen is a focused, high-ownership team. We value clear thinking, direct communication, careful execution, and people who carry important work through to a result. We expect colleagues to surface inconvenient evidence, ask precise questions, and make the limits of their work explicit. Strong specialist judgment matters, as does the willingness to understand how your work affects research, markets, engineering, and risk. We prefer useful documentation and clear decisions to unnecessary process.