Aaron Ontoyin Yin

Anirudh Jaidev Mahesh, Ben Griffin, Fuat Alican et al.

Large language models (LLMs) are increasingly used for high-stakes decision-making, yet existing approaches struggle to reconcile scalability, interpretability, and reproducibility. Black-box models obscure their reasoning, while recent LLM-based rule systems rely on per-sample evaluation, causing costs to scale with dataset size and introducing stochastic, hallucination-prone outputs. We propose reframing LLMs as code generators rather than per-instance evaluators. A single LLM call generates executable, human-readable decision logic that runs deterministically over structured data, eliminating per-sample LLM queries while enabling reproducible and auditable predictions. We combine code generation with automated statistical validation using precision lift, binomial significance testing, and coverage filtering, and apply cluster-based gap analysis to iteratively refine decision logic without human annotation. We instantiate this framework in venture capital founder screening, a rare-event prediction task with strong interpretability requirements. On VCBench, a benchmark of 4,500 founders with a 9% base success rate, our approach achieves 37.5% precision and an F0.5 score of 25.0%, outperforming GPT-4o (at 30.0% precision and an F0.5 score of 25.7%) while maintaining full interpretability. Each prediction traces to executable rules over human-readable attributes, demonstrating verifiable and interpretable LLM-based decision-making in practice.

Maximilian Westermann, Ben Griffin, Aaron Ontoyin Yin et al.

Feature discovery from complex unstructured data is fundamentally a reasoning problem: it requires identifying abstractions that are predictive of a target outcome while avoiding leakage, proxies, and post-outcome signals. With the introduction of ever-improving Large Language Models (LLMs), our method provides a structured method for addressing this challenge. LLMs are well suited for this task by being able to process large amounts of information, but unconstrained feature generation can lead to weak features. In this work, we study reasoning control in LLMs by inducing cognitive behaviors for improving feature discovery. We introduce CoFEE (Cognitive Feature Engineering Engine), a reasoning control framework that enforces cognitive behaviors in how the LLM reasons during feature discovery. From a machine learning perspective, these cognitive behaviors act as structured inductive biases over the space of candidate features generated by the model. These behaviors have been exploited with success in ML models, and include backward chaining from outcomes, subgoal decomposition, verification against observability and leakage criteria, and explicit backtracking of rejected reasoning paths. In a controlled comparison, we show that enforcing cognitive behaviors yields features with higher empirical predictability than those under unconstrained vanilla LLM prompts. CoFEE achieves an average Success Rate Score that is 15.2% higher than the vanilla approach, while generating 29% fewer features and reducing costs by 53.3%. Using held-out feature evaluation, we assess whether cognitively induced features generalize beyond the data used for discovery. Our results indicate that, in our evaluated setting, reasoning control is associated with improvements in quality and efficiency of LLM-based feature discovery.

Sichao Xiong, Yigit Ihlamur, Fuat Alican et al.

Traditional decision tree algorithms are explainable but struggle with non-linear, high-dimensional data, limiting its applicability in complex decision-making. Neural networks excel at capturing complex patterns but sacrifice explainability in the process. In this work, we present GPTree, a novel framework combining explainability of decision trees with the advanced reasoning capabilities of LLMs. GPTree eliminates the need for feature engineering and prompt chaining, requiring only a task-specific prompt and leveraging a tree-based structure to dynamically split samples. We also introduce an expert-in-the-loop feedback mechanism to further enhance performance by enabling human intervention to refine and rebuild decision paths, emphasizing the harmony between human expertise and machine intelligence. Our decision tree achieved a 7.8% precision rate for identifying "unicorn" startups at the inception stage of a startup, surpassing gpt-4o with few-shot learning as well as the best human decision-makers (3.1% to 5.6%).

Te Pei, Fuat Alican, Aaron Ontoyin Yin et al.

This paper introduces GPT-HTree, a framework combining hierarchical clustering, decision trees, and large language models (LLMs) to address this challenge. By leveraging hierarchical clustering to segment individuals based on salient features, resampling techniques to balance class distributions, and decision trees to tailor classification paths within each cluster, GPT-HTree ensures both accuracy and interpretability. LLMs enhance the framework by generating human-readable cluster descriptions, bridging quantitative analysis with actionable insights.

Rick Chen, Joseph Ternasky, Aaron Ontoyin Yin et al.

Large language models (LLMs) can already identify patterns and reason effectively, yet their variable accuracy hampers adoption in high-stakes decision-making applications. In this paper, we study this issue from a venture capital perspective by predicting idea-stage startup success based on founder traits. (i) To build a reliable prediction model, we introduce LLM-AR, a pipeline inspired by neural-symbolic systems that distils LLM-generated heuristics into probabilistic rules executed by the ProbLog automated-reasoning engine. (ii) An iterative policy-evolution loop incorporates association-rule mining to progressively refine the prediction rules. On unseen folds, LLM-AR achieves 59.5% precision and 8.7% recall, 5.9x the random baseline precision, while exposing every decision path for human inspection. The framework is interpretable and tunable via hyperparameters, showing promise to extend into other domains.

Rick Chen, Joseph Ternasky, Afriyie Samuel Kwesi et al.

Benchmarks such as SWE-bench and ARC-AGI demonstrate how shared datasets accelerate progress toward artificial general intelligence (AGI). We introduce VCBench, the first benchmark for predicting founder success in venture capital (VC), a domain where signals are sparse, outcomes are uncertain, and even top investors perform modestly. At inception, the market index achieves a precision of 1.9%. Y Combinator outperforms the index by a factor of 1.7x, while tier-1 firms are 2.9x better. VCBench provides 9,000 anonymized founder profiles, standardized to preserve predictive features while resisting identity leakage, with adversarial tests showing more than 90% reduction in re-identification risk. We evaluate nine state-of-the-art large language models (LLMs). DeepSeek-V3 delivers over six times the baseline precision, GPT-4o achieves the highest F0.5, and most models surpass human benchmarks. Designed as a public and evolving resource available at vcbench.com, VCBench establishes a community-driven standard for reproducible and privacy-preserving evaluation of AGI in early-stage venture forecasting.

Mihir Kumar, Aaron Ontoyin Yin, Zakari Salifu et al.

This paper presents a framework for predicting rare, high-impact outcomes by integrating large language models (LLMs) with a multi-model machine learning (ML) architecture. The approach combines the predictive strength of black-box models with the interpretability required for reliable decision-making. We use LLM-powered feature engineering to extract and synthesize complex signals from unstructured data, which are then processed within a layered ensemble of models including XGBoost, Random Forest, and Linear Regression. The ensemble first produces a continuous estimate of success likelihood, which is then thresholded to produce a binary rare-event prediction. We apply this framework to the domain of Venture Capital (VC), where investors must evaluate startups with limited and noisy early-stage data. The empirical results show strong performance: the model achieves precision between 9.8X and 11.1X the random classifier baseline in three independent test subsets. Feature sensitivity analysis further reveals interpretable success drivers: the startup's category list accounts for 15.6% of predictive influence, followed by the number of founders, while education level and domain expertise contribute smaller yet consistent effects.