Rapheal Obodugo

Rapheal Obodugo

Lagos · Riyadh · London/
THE THESIS

Most financial software assumes numbers are correct because an API returned them. We build deterministic multi-agent architectures that prove they are.

2026 — Present
AvaeFounder & Applied Researcher

Continuous audit for modern finance. Researching deterministic multi-agent verification architectures that audit 50,000+ transaction batches without hallucination.

2024 — 2025
MySubFounder · Exited

Built consumer recurring payment infrastructure, scaled to 360,000 accounts across emerging markets, and successfully completed acquisition.

2022 — 2024
Derayah FinanceHead of Engineering

Led core engineering across digital wealth management, capital market execution engines, and retail investment platforms in Riyadh.

2021 — 2022
MetaInfrastructure Engineering

Distributed systems infrastructure, large-scale data pipelines, and network reliability engineering.

2026 — Present
OriafoDirector

A commercial palm kernel processing mill in Bende, Abia State, Nigeria, commissioning industrial processing equipment for vegetable oil production.

SSRN Electronic Preprints · Financial ForensicsSSRN ID 6944801

Adversarial Consensus Verification for Reliable LLM agents in Financial Forensics: The Pave Interchange Benchmark

Zacchaeus Bolaji (Harvard)·Rapheal Obodugo (Avae / Imperial College London)·John Hella (Imperial / Avae)
The Problem

Single-agent LLMs degrade below 55% accuracy on forensic discovery tasks due to relational blindness, numerical drift, and contextual overload.

The Benchmark Result

Avae 2.0 multi-agent adversarial consensus achieves 100% forensic accuracy across all 24 tasks on a 50,000-record payment reconciliation suite.

I am building Avae, an applied research and engineering company focused on continuous audit for modern finance. We build systems that can reliably reason over financial data, ensuring that numbers are not merely available, but verifiably correct.

Modern financial workflows still rely heavily on manual reconciliation, disparate spreadsheets, and brittle logic. As these systems scale, ensuring absolute correctness becomes increasingly difficult. Avae exists to solve this problem at its root by building deterministic systems capable of processing large-scale financial data, detecting structural inconsistencies, and explaining outcomes with absolute precision.

I spend a lot of time reading mostly history, philosophy, and biographies. I enjoy understanding how exceptional people thought and why complex systems work the way they do. Occasionally, I turn to fiction when I need the space to escape.

I write about the mechanics of building products, the art of execution, and the unique lessons that come from operating in emerging markets.

Outside Avae, I'm a director of Oriafo, a palm kernel processing mill in Abia State, Nigeria.

I also advise early-stage startups on product strategy and engineering. Get in touch if you are building something genuinely interesting.

All 20 essays →
Full Bookshelf →
How AI is reshaping the future of work & skills in AfricaCNBC Africa

How AI is reshaping the future of work & skills in Africa

Broadcast Interview · Africa Market Strategy
The Future of Work: The 9-to-5 and the Gig Economy in the Age of AIPanel

The Future of Work: The 9-to-5 and the Gig Economy in the Age of AI

Panel Discussion · Generative Systems
Stevie Award (Gold)Achievement in Science or Technology (Middle East & Africa)