3. Data, Science & AI

Deterministically constructs AI agent prompts using Python, enabling structured input for BAZINGA agents prior to deployment.

Context assembler for AI agents: prioritizes and ranks relevant information, enforces token budgets, and captures error patterns to optimize agent performance.

Manages agent interactions, reasoning logs, and token tracking for AI systems, providing visibility into agent behavior and resource usage.

Provides a unified dashboard for aggregating and visualizing project quality metrics in real-time.

Manages session lifecycles, state saving, dashboard data access, and system queries through database operations.

Validates data integrity and state consistency during Google Sheets operations to ensure reliable data processing.

Type-safe processing of outputs from Loop 2 validators and Loop 3 implementers, handling agent confidence scores and consensus calculation.

Automates structured persistence of agent outputs and confidence scores in SQLite for tracking across CFN Loop workflows.

Enables automatic data persistence for CFN Loop using SQLite storage and Redis coordination.

Manages AI agent lifecycle from selection through completion, including spawning, execution, output processing, and audit trails.

Selects optimal AI agents for tasks using classification, with automatic fallback for reliability in AI workflows.

Classifies task descriptions into predefined categories to enable automated agent selection in multi-agent systems.

Proactively retrieves relevant past user insights from memory to enhance response quality without explicit requests.

CLI tool for Are.na enabling incremental block exports, AI-generated image enrichment, and browsable HTML views for content management.

Builds resilient pipelines for incremental data ingestion from paginated APIs, handling rate limits, duplicates, and historical backfills.

Expert guidance for building, optimizing, and debugging PostgreSQL 16 SQL queries and managing database schemas.

Provides real-time speech-to-text transcription and audio analysis via AssemblyAI's streaming APIs for live meetings and voice agents.

Extracts text from EPUB, MOBI, and PDF ebooks, converting them to plain text for analysis and processing.

Designs database architectures for data-heavy applications, optimizing schema, indexing, and scaling strategies while addressing performance issues.

Provides guidance on agent coordination patterns, communication protocols, and report formats for parallel agent workflows within the sprint system.

Efficiently runs and edits Jupyter notebooks with headless execution, using uv for dependency management.

Identifies true causes by distinguishing correlation from causation, testing hypotheses, and controlling for confounding variables in data analysis.

Empirically validates symmetries in data and machine learning models, providing test protocols and metrics for invariance and equivariance checks.

Optimizes AI prompt structures to ensure consistent, reliable, and safe outputs through explicit constraints, domain encoding, and multi-step reasoning decomposition.