About
The independent R&D program of Daniel Higuera, run since 2017 in parallel with industrial R&D leadership.
AES Research — Applied Emergent Sciences — is the independent AI research vehicle of Daniel Higuera. The work started in 2017 as a structured program for exploring production AI architecture outside the constraints of any employer’s roadmap. It has run continuously since — through a full-time deep-focus sprint from 2020 to 2022, and alongside industrial R&D leadership at Hitachi Energy from 2022 to present.
The focus is agentic systems: the architectural patterns that let multi-agent systems operate reliably over long horizons, across many sessions, in real production conditions. Not benchmark performance. Not demo quality. The problems that show up when context compounds, memory has to persist, autonomous outputs have to be verifiable, and the human can’t be in the loop for every decision.
The current body of work spans 12+ AI-native projects under a unified director-agent architecture. Two-tier memory, skeptic membranes, role-tagged sub-agent dispatch, and UAT harnesses for residual context gap detection — these are all patterns that emerged from real failures in real systems, not from reading about them.
Industrial track
Twenty years of professional experience across energy markets, production ML, and industrial software. Eleven years of direct-report management.
R&D Manager, Hitachi Energy (2022 – present). Owns PROMOD — the industry-standard grid-planning and wholesale-market simulation platform used by North American ISOs and utilities. Previously owned the Bid-to-Bill wholesale-settlement system before pivoting to current scope.
Data Platform Owner, Innowatts (2019 – 2020). Managed a 10–15-person globally distributed engineering team across the US, India, Ireland, England, and Spain. Big-data ETL, SOC2 + ISO 27001 compliance, SaaS architecture, novel ML algorithms for utility-customer segmentation.
Manager, Market Sync & Invoice Operations, NRG Energy (2012 – 2017). Ten direct reports owning 810 transactions, the Market Data Repository, the Low Income program, QC, invoice audit, and project management.
Reliant Energy (2006 – 2012). Six-year progression from ETM Rep through Technical Supervisor / Analyst III owning 810s and Market Sync.
AI and ML developed independently over 15+ years, predating formal academic curricula for most of the sub-disciplines the work now covers. The energy-market background is not incidental — the same calibration discipline that keeps a grid-planning model honest translates directly to keeping an autonomous agent honest.
Selected work
- An AI-native investment research platform with 45+ analytical tabs, a multi-agent research pipeline running parallel specialist agents, and an integrated portfolio decision engine with risk decomposition.
- A dynamic human-behavior simulation engine driving an interactive narrative experience — characters with persistent state, autonomous goal-directed behavior, and consequence-tracked plot arcs.
- A cross-project director-agent architecture coordinating 12+ active projects via role-tagged sub-agent dispatch, two-tier memory, and daily pipeline consolidation — the system that produced and manages the work described on this site.
- A production ML pipeline with factor capture, ensemble clustering, regime labeling, and multi-year out-of-sample validation across equity market data.
- A context-rich children’s learning platform with 7+ playable game prototypes and a voice-driven guide agent designed around COPPA constraints and pre-reader UX.