My Research

My research is culmination of a process that begins with the journey of an idea, from a simple "what if" to a tangible impact on the world. This is the "Entelechy" mission that drives my work.

Curiosity

It always begins with curiosity. This is the spark a "what if" question that challenges an assumption.

Inquiry

Curiosity becomes inquiry. This is the deep dive, the research. It's about asking the foundational questions to find the real problem, guided by the "Sankofa" principle of looking to established wisdom to build a wiser future.

Discovery

Inquiry leads to discovery. This is core truth that unravels a new framework or potential.

Application

Application is where I move from researcher to builder. It's the process of turning that new truth into a stable, well-engineered, and useful product that people can actually touch.

Abundance

This is the ultimate goal: the realization of potential. It's the outcome the product has on its users and the world. This is the "why."

Research

Research Foundation

Neuro-Symbolic AI and Causal Inference: My Approach to Legal Reasoning

How I'm combining neural pattern recognition with symbolic causal reasoning to build "glass box" legal AI systems. This article explains the technical foundation of my MS thesis work at Cornell bridging interpretability with systematic reasoning through attention analysis and causal mapping.

Read full article 15 min read

MS Thesis / Deployed Product

Search-Augmented Reinforcement Learning for Legal Reasoning

MS Thesis, Cornell 2025 | Advisor: Claire Cardie | Defended September 5, 2025

First framework integrating legal database access during RL training episodes (not just inference).
Multi-task evaluation with jurisdictional compliance across all US legal systems.
10.5pp improvement on LegalBench through tool-assisted training with GRPO.
Real-World Impact: This thesis is the core of LexLens, a deployed product that helps legal professionals build arguments with 100% verifiable sources.

Working Paper (coming soon) Code

Articles

Fundamentals

Understanding Reasoning in Language Models

An introduction to how reasoning models work from Chain-of-Thought to Tree-of-Thought approaches. Covers the fundamentals of o1, o3, and DeepSeek R1, along with current limitations including hallucination rates (33% for o3) that my research aims to address.

News

Product Strategy

On Pause: Why We’re Iterating on LexLens

Shipping a product is the beginning of the journey, not the end.

After launching the first version of LexLens which successfully deployed my Cornell thesis research we did the most important part of the product cycle: we listened. We gathered invaluable feedback from the lawyers and students using the app, and they showed us a deeper, more complex problem than the one we had initially set out to solve.

That's why LexLens.app is currently paused. This isn't a stop; it's a deliberate, iteration to solve the problem that actually matters. Meridius, is now helping guide the re-architecture of the core product based on this user wisdom.

Want to discuss these ideas?

I'm always happy to chat about AI reasoning, interpretability, and building trustworthy systems.

Get in Touch