Project Overview
Bayou Analytica is a Louisiana K-12 decision-support platform that transforms how education leaders and policymakers discover, simulate, and fund interventions across the state’s 64 parishes. Built for the NexusLA DevDays EdTech challenge, it combines a FastAPI backend, a React 19 + TypeScript frontend, and a massive public-data ingestion pipeline to surface actionable intelligence — from parish-level health scores to budget-optimized investment portfolios.
1st Place (Solo) — Nexus Louisiana DevDays 2026
The platform is currently under review by Louisiana state policymakers as a potential tool for data-driven education investment planning.
Key Features
Interactive Map Dashboard
A choropleth map of all 64 Louisiana parishes with 10+ metric layers, a year slider spanning 2007–2025, time-lapse animation, and drill-down capabilities for detailed parish analysis.
Funder’s Lens (/funder)
A budget-constrained marginal-ROI portfolio optimizer. Enter a budget ($500K–$20M) and receive a ranked investment portfolio across parishes and innovation types, sorted by graduates per $100K, complete with “Do Not Build” recommendations.
Policy Simulator
What-if simulation across 11 policy levers — from teacher salary to literacy investment — with empirical elasticities derived from Louisiana parish panel data spanning 20+ years.
Allie AI Advisor
A natural-language investment analyst powered by an LLM agent with 28 tools (17 data analysis, 11 UI render), streaming NDJSON events, and a Fastlane deterministic mode for common funder questions with zero LLM latency.
Innovation Matchmaker & Zones
Scores 8 school-model types per parish (literacy intervention, manufacturing CTE, teacher pipeline, etc.) and clusters parishes by workforce similarity for smarter pilot programs.
Workforce Alignment
Maps credential pipelines against industry demand and tracks major economic development projects (Meta, Woodside, NASA) across parishes.
Data Transparency
Full data provenance — every source file is browsable, every parser step auditable, and every AI conversation logged. 35 automated QA checks run on every data import.
Technologies Used
- Backend: FastAPI, Pydantic, pandas/polars, DeepSeek, PydanticAI
- Frontend: Vite, React 19, TypeScript, Tailwind CSS v4, shadcn/ui
- Map:
react-map-gl+ Mapbox GL JS - AI Agent: 28-tool LLM agent with streaming NDJSON events
- Data Pipeline: Louisiana education datasets parsed to Parquet and aggregated by parish/year
- Infrastructure: Docker, Railway
Technical Highlights
- 28-tool LLM agent and greedy optimizer evaluating 512 investment candidates via a multi-model ensemble across 1,235 schools
- Sub-50ms in-memory query layer processing 1,286+ datasets via an OpenAPI pipeline auto-generating React 19 TypeScript types
- Empirical elasticities derived from Louisiana parish panel data for 11 policy levers with cost-per-outcome estimates
- Full data provenance with 35 QA checks on every import
The Origin Story
Bayou Analytica was built for the Nexus Louisiana DevDays EdTech challenge — a statewide innovation competition for Louisiana college students. The challenge: build a tool that helps education leaders make better, data-informed decisions about K-12 investments across Louisiana.
Over an intensive build period, the platform grew from a concept to a full-stack, deployment-ready application with 1,286+ ingested datasets, an AI advisor, policy simulator, and budget optimizer. It won 1st Place (Solo) at the competition and is now under review by state policymakers.
Project Status
Bayou Analytica is actively maintained and available at www.bayouanalyti.ca. The platform is free to use and built for Louisiana’s students, educators, and policymakers.
Screenshots
