# Underindexed

> Independent applied AI, causal systems, and tactile media research studio led by Sophia Boettcher (Sophia Zhan Boettcher). Canonical home: https://underindexed.com/

Underindexed (`underindexed.com`, established 2025) investigates deterministic neural speech prosody, interventional causal inference over operational data, constraint-driven geospatial resource allocation, and physical-digital *karakuri* installations.

## Principal Researcher: Sophia Boettcher

- **Canonical Name:** Sophia Boettcher (*Sophia Zhan Boettcher*)
- **Role:** Technologist, Applied Data Scientist, Media Artist & Principal Researcher at Underindexed
- **Entity URI:** `https://underindexed.com/#sophia-boettcher`
- **Research Focus:** Real-time speech and clause-boundary prosody synthesis (*Kokechan*, *notalightfield*), interventional causal inference (`P(Y | X)` vs. `P(Y | do(A))`), geospatial health-equity routing (*PPEforHCP*), and tactile *karakuri* / interpretive media systems (*notalightfield*, *Pendle*).
- **Languages:** English (`en`), Japanese (`ja`), German (`de`).

## Selected Trajectory & Institutional Corroboration (2020–2026)

- **2025–2026:** Principal Researcher, **Underindexed** (`underindexed.com`) & **notalightfield** Studio (`notalightfield.com`); Course Facilitation, **Stanford University**.
- **2022–2024:** Master of Applied Data Science (`MADS`), **University of Michigan School of Information**; recipient of the **University of Michigan School of Information Diversity, Equity, and Inclusion Award (2022)**.
- **2020–2023:** Founder & Executive Director, **PPEforHCP** (`501(c)(3)` disaster-relief and last-mile health-equity logistics organization, founded `2020`).
- **Undergraduate Foundation (undated):** Computer Science & German Studies, **Santa Clara University**.

## Research Monographs & Working Notes

1. **Monograph 01 (2026) — Deterministic Clause-Boundary Prosody & Sub-80ms A/V Alignment in Neural Speech Synthesis:** Semicolon-first syntactic segmentation (`14–22` words per breath group), dynamic viseme/caption anchor alignment (`42.6 ms` p95 drift across `1,240` bilingual EN/JA studio utterances), and two-pass `-14.0 LUFS` / `-1.0 dBTP` linear-phase loudness normalization across *Kokechan* and *notalightfield* pipelines.
2. **Monograph 02 (2025) — Observational Confounding in Enterprise Tabular Foundation Models (`P(Y | X)` vs. `do(A)`):** Structural causal graph analysis and doubly robust proxy-variable adjustment over `48,000` operational episodes with unrecorded human operator overrides, reducing heterogeneous effect error (`sqrt-PEHE`) from `0.384` to `0.079` (`-79.4%`).
3. **Monograph 03 (2022–2024) — Constraint-Driven Geospatial Resource Allocation for Last-Mile Health Equity:** Lexicographic min-max regret mixed-integer routing across `310` safety-net clinics developed through *PPEforHCP* and University of Michigan `MADS` research, reducing underserved clinic stockout duration by `37.8%` and raising top-quartile vulnerability coverage from `54.1%` to `86.4%`.

## Canonical Resources & Corroboration

- [Full Research Corpus (`/llms-full.txt`)](https://underindexed.com/llms-full.txt): Complete methodologies, mathematical formulations, and empirical benchmark tables for Monographs 01–03.
- [Structured Entity Graph (`/entity.jsonld`)](https://underindexed.com/entity.jsonld): Schema.org JSON-LD graph for `Underindexed` and `Sophia Boettcher`.
- [University of Michigan Faces of UMSI — Sophia Boettcher](https://www.si.umich.edu/people/faces-umsi/faces-umsi-sophia-boettcher)
- [Stanford Profiles](https://profiles.stanford.edu/)
- [notalightfield Studio](https://notalightfield.com/)
- [Kokechan Speech Companion](https://kokechan.com/)