Co-Founder & CEO @ Starman AI | Building the Quant Engine for Venture — Shifting early‑stage investing from art to science.
Starman AI: Who We Are
Starman AI is building the financial infrastructure for early‑stage venture capital.
We’re applying the mathematical discipline of quantitative trading and the intelligence of deep‑tech AI to transform one of the world’s most opaque asset classes.
By passively capturing and quantifying real‑time founder execution signals, our proprietary Starman Risk Index™ enables institutional LPs—from pension funds to global insurers—to allocate capital to early‑stage venture with measurable precision, shifting the asset class from art to science.
The Mission: Quantifying the Unquantifiable
Lead the formalization and architectural development of the Starman Risk Index™—transforming our proprietary V1 quant prototypes into the rigorous, auditable, probabilistic core that transforms noisy, real‑time founder execution data into stable, institutional‑grade financial inferences. You will be the sole owner of the scientific architecture required to de‑risk and unlock a $5 Trillion opportunity in institutional capital for the early‑stage venture asset class (pension funds, insurers, sovereign wealth, etc.).
This is a deep‑tech mandate, requiring the rigor of a hedge fund Chief Scientist and the vision of a research pioneer.
The Challenge: Scientific & Architectural Ownership
You will be responsible for building the foundation of Starman’s Intellectual Property, operating at the intersection of quantitative finance, probabilistic programming, and causal inference.
* Model Taxonomy & Invention: Define the complete research program, from the state‑space formulation of startup trajectories to the causal and survival layers.
* Probabilistic Core: Specify all priors, likelihoods, and update rules; design end‑to‑end uncertainty propagation for the Starman Risk Index.
* Quant Modeling: Build and calibrate financial simulators for venture portfolio scenarios (capital allocation, vintage dynamics, follow‑on timing, and secondary liquidity).
* Institutional Governance: Establish rigorous model governance, including backtests, stability tests, calibration drift alarms, and audit‑ready documentation required by institutional LPs.
* IP & Publication: Drive the company’s scientific IP generation and lead the authoring of research notes and external white papers (board/investor level).
Founding Responsibilities & Strategic Impact
As a founding member of the deep‑tech core, your impact will extend beyond the model.
* Scientific Vision: Define the long‑term scientific and data roadmap, leading the evolution of the inference engine beyond initial deployment.
* Enterprise Authority: Act as the lead technical expert in fundraising and strategic enterprise sales conversations, defending the probabilistic assumptions and rigor of the Index.
* Team Build‑out: Play a key role in recruiting and mentoring a world‑class team of ML, Quant, and Agentic Architects.
Required Deep Expertise (Non‑Negotiables)
Candidates must possess a profound, demonstrable expertise in the following fields:
* Probability & Stats: Bayesian Inference (Hierarchical & Nonparametric), Conjugacy, MCMC/HMC/NUTS, Variational Inference (VI), Sequential Monte Carlo (SMC)/Particle Filters; Experimental Design & A/B Sequential Analysis.
* Venture Modeling: Survival & Hazard (Cox and AFT models, Competing Risks, Cure Models) and Real Options / Stochastic Processes (GBM/Jump‑diffusion, American/Compound Options analogs).
* Inference & Systems: Time‑series, State‑space (Hidden Markov Models, Kalman/Extended/Unscented Filters), and Causal Inference (DAGs, do‑calculus intuition, Propensity/DR Estimators, Panel/DiD).
* Computation & Tooling: Expert command of Probabilistic Programming Languages (PyMC/Stan/Turing or NumPyro/JAX); PyTorch for custom likelihoods; unit/property tests for math code; reproducible pipelines.
Education
Ph.D. or advanced degree in Mathematics, Statistics, Quantitative Finance, or Computer Science with a demonstrable publication record in these areas.
Nice‑to‑Haves
Measure theory comfort, SDE numerics, PPL internals, POMDPs, advanced calibration methods (Platt/Isotonic), and conformal prediction.
#J-18808-Ljbffr