# jiokua ai builder/policy nerd Studying agent-mediated human coordination in the wild. Exploring how human-to-agent-to-human workflows reshape trust, labor, and social commitment. ## Contact Website: https://jiokua.dev Email: hello@jiokua.dev ## Experience ### intentlii — Founder & CEO Feb 2026 – Jun 2026 · Oakland, CA A deployed iOS testbed for agent-mediated human coordination — studying how AI in interpersonal commitments shifts trust, attribution, and delegation. - Designed an "agents inform but do not commit" pattern that lets AI surface structured suggestions while preserving human authorship over social commitments — the foundation of the product's trust model. - Built an MCP server exposing intent and coordination primitives to third-party agents, positioning intentlii as agent-ready coordination infrastructure. - Architected and shipped the full stack solo: React Native/Expo client, GCP backend (Cloud Run, Cloud SQL/PostgreSQL, GCS, Pub/Sub), TypeScript monorepo. - Owned product, engineering, design, and go-to-market end-to-end; in-cohort pivot within Berkeley SkyDeck Batch 20. ### Gheek AI — Founder & CEO Jan 2025 – Feb 2026 · Oakland, CA An end-to-end customer-engagement platform for building assistive chatbots over a business's own pages, documents, and forms. - Admitted to Berkeley SkyDeck Batch 20 (2025); shipped to real customers, including photographer Sapna Reddy, who turned workshop brochures into chat-driven experiences. - Architected a multi-tenant GCP stack (Cloud Run, Cloud SQL/PostgreSQL, GCS/Cloud CDN, Pub/Sub) with a TypeScript monorepo spanning Next.js web and Expo mobile. - Designed analytics-ready data infrastructure (event modeling, GIN/trigram indexing) and guided-support UX (snippet capture, streaming responses, bookmarking). - Managed a small team through scoped milestones and releases; pivoted the underlying coordination insights into intentlii. ### Fimio, Inc. — Founding Machine Learning & Software Engineer Jun 2022 – Jul 2024 · San Francisco, CA Founding engineer building ML tooling for smart-contract auditing and reproducible Python environments. - Built ML-powered Solidity smart-contract auditing — LLM analysis with retrieval over known vulnerability patterns — to surface security issues for auditors. - Built ML-driven reproducibility tooling that recreates working Python environments from existing code and dependency artifacts. - Shipped scalable AWS microservices (containerized services, data pipelines, APIs) and content-addressed data infrastructure for provenance, dedup, and faster retrieval. - Owned end-to-end ML — datasets, evaluations, and production models — across both products. ### Madojo — Co-Founder & Product Lead Oct 2021 – Jul 2022 · Berkeley, CA Co-founded a platform connecting Nigerian graduates to tech employers through skills training and mentorship. - Won the 2022 Binance-LIFT "Blockchain for Social Good" Grand Prize. ### CITRIS and the Banatao Institute, UC Berkeley — AI Policy Researcher Jul 2021 – Jan 2022 · Berkeley, CA AI-governance research with the CA Department of Technology on EU AI Act analysis. - Collaborated with the CA Department of Technology and former CA Chief Data Officer Joy Bonaguro on EU AI Act analysis to inform California's emerging AI governance. - Co-developed an AI risk-management curriculum delivered to city, county, and state agencies across California. ### Radiant Earth Foundation — Data Policy Researcher Jun 2021 – Nov 2021 · Washington, D.C. (remote) Researched agricultural data-ownership frameworks and data ethics for ML on satellite imagery. - Authored a white paper on agricultural data ownership across high- and low/middle-income contexts; published as part of the Gates Foundation's ECAAS initiative. - Researched data ethics in satellite imagery and ML, focused on equity implications for the developing world. ### Blum Center for Developing Economies, UC Berkeley — Industry 4.0 Policy Consultant Jan 2021 – May 2021 · Berkeley, CA Led Future-of-Work research with former Dep. Asst. Treasury Secretary Brad DeLong. - Led a research project with Brad DeLong on COVID-19's effects on Vietnamese and Mexican manufacturing and the Future of Work. - Translated findings into policy recommendations for low- and middle-income economies facing accelerated automation. ### University of California, Berkeley — Graduate Student Instructor & Reader Fall 2020 – Spring 2022 · Berkeley, CA Taught three graduate courses in public policy and data/information visualization. - Taught and assisted in Public Policy Implementation, Data Visualization in Public Policy (Goldman School), and Information Visualization (School of Information). - Mentored students in coding best practices, ethical data handling, and applied analysis. ### Goldman School of Public Policy — Global Programs Specialist Jul 2017 – Oct 2019 · Berkeley, CA Ran executive-education programs for foreign government officials and corporate leaders. - Developed and coordinated short- and long-term executive-education programs for foreign officials, university leadership, and corporate executives. - Owned program operations end-to-end: contracts, client research, logistics, and web maintenance. ## Education ### Master of Development Engineering — AI / Data Analytics for Social Impact University of California, Berkeley · 2022 · 3.72 / 4.00 Coursework: - CS 188 — Artificial Intelligence - CS 288 — Natural Language Processing - CS 294 — Algorithms and Inequality - INFO 251 — Applied Machine Learning - PUBPOL 210A/B — Economic Public Policy Analysis - PUBPOL 240A/B — Decision Analysis & Quantitative Methods - PUBPOL 271 — Political Economy - PUBPOL 273 — Public Management ### Master of Public Policy — Graduate Certificate in Applied Data Science University of California, Berkeley · 2021 ### B.A. Political Science University of California, Berkeley · 2017 ### B.A. English University of California, Berkeley · 2017 ## Skills - Full-stack TypeScript - React Native / Expo - ML & LLM systems - Agent systems & MCP - GCP / AWS infrastructure - Eval & data infrastructure - AI policy & governance - Product strategy ## Publications ### Manufacturing During the Pandemic: COVID-19's Impact on Mexico and Viet Nam Author · Blum Center for Developing Economies · 2021 A field-research-informed study of pandemic-era manufacturing restructuring and the Future of Work in two middle-income economies pursuing post-China supply-chain reorientation. ### Big Data in Agriculture: Achieving Data Ownership for Farmers Across Regions Lead Researcher · Radiant Earth Foundation · Gates Foundation ECAAS · 2021 A white paper on agricultural data-ownership frameworks across high- and low/middle-income contexts, published institutionally as part of the Gates Foundation's ECAAS initiative. ### Designing New Ways to Build Worker Power Co-Author · Jobs for the Future · 2020 A Goldman School policy report on state-level interventions to build worker power in California — sectoral bargaining, co-enforcement, employee ownership, and procurement. ### Vietnam and the Consequences of Technological Innovation for Developing Countries Author · Center for Strategic & International Studies · 2017 Examined automation's distributional effects on developing economies through the lens of Vietnam's manufacturing sector. ## Selected Projects ### Clide A product surface for reading, shaping, and acting on complex work. Role: Designer and builder Category: Product Technologies: - Product strategy - Interface design - Frontend Highlights: - Clarified a complex product surface for faster review. - Connected release notes, source reading, and agent workflows into one legible portfolio thread. ### Cheesy Theseus An agent-legibility eval for non-code research workspaces: testing how structure, retrieval, verification, and navigation change whether a cold agent stays grounded or abstains. Role: Researcher and builder Category: Research Technologies: - Python - Eval design - Agent tooling - Deterministic graders Highlights: - Built H-line eval verticals that isolate verification, retrieval, structure, navigation, and familiarity over controlled workspace surfaces. - Scored runs with deterministic graders and a 2x3 sufficiency, confabulation, and abstention grid. - Tracked capability-dependent findings around when affordances help answerability and when they hurt calibrated abstention. ### Redpen A personal capture tool that turns photographed red-pen book annotations into markdown highlights, with an iPhone capture app, deployable FastAPI pipeline, and text-first library. Role: Designer and builder Category: Product Technologies: - Python - FastAPI - SwiftUI - Azure AI Vision Read - Azure Container Apps Highlights: - Built a deterministic red-ink mask, connected-component classifier, OCR matcher, and markdown export path for annotated book pages. - Shipped an iPhone capture flow that uploads full-resolution page photos and returns extracted highlights with an annotated overlay. - Moved captures into a text-first library where markdown and metadata persist while source photos remain disposable cache artifacts.