3. Data, Science & AI

Generates optimized AI prompts across portrait, cross-domain, and design contexts using semantic understanding and consistency checks.

Guides database schema modifications including table/column additions, relations, and data access layer updates for schema migrations.

Provides optimized Postgres query patterns, schema design guidelines, and configuration tips based on Supabase's database best practices for performance and scalability.

SDK for scalable LLM evaluation across 100+ benchmarks with containerized execution on local, HPC, and cloud platforms.

Migrates memory blocks between AI agents for sharing, inheritance, or initialization during agent setup.

Enables communication between software agents on a server for task delegation, querying, and coordination in multi-agent systems.

Provides a guide for adding new LLM models to Letta Code, covering configuration, testing, and validation processes.

Generates statistical charts including bar, line, scatter, and radar visualizations from numeric data arrays using Vega-Lite and Vega.

Creates complex directed and undirected graphs with precise layout control for dependency trees, network topologies, and module relationships.

Extracts text from images using the Tesseract OCR engine, enabling conversion of visual content into editable text data.

Transcribes audio files to text using the OpenAI Whisper speech recognition model.

Performs basic operations on user-provided text files, including content summarization for data insights.

Enables asynchronous delegation of tasks to Codex AI model, ending the current turn immediately upon delegation.

Delegates user tasks to Gemini AI via CLI, awaiting responses. Only for explicit user delegation to Gemini, not for Gemini-related queries.

Verifies connectivity to Google's Gemini AI model via CLI, enabling status checks and troubleshooting for AI service interactions.

Enables asynchronous data processing using Dask, activated exclusively through explicit user delegation to the Droid agent.

Delegates user tasks to Droid via CLI and returns the response.

Fetches the latest response from Google's Gemini AI model via CLI command, triggered only by explicit user request.

Enables asynchronous delegation to Gemini AI model with immediate turn termination upon user command.

Executes bulk RNA-seq differential expression analysis using omicverse, including DESeq2 normalization, statistical testing, and pathway enrichment.

Preprocesses TCGA bulk RNA-seq data including expression, clinical, and sample metadata using omicverse, generating annotated AnnData files for downstream analysis.

Performs statistical tests, hypothesis testing, and correlation analysis using scipy and statsmodels for data-driven insights.

Maps scRNA-seq atlases to spatial transcriptomics slides using deep-forest AI training and marker visualization for genomic analysis.

Converts bulk RNA-seq data into synthetic single-cell datasets using Bulk2Single workflow for cell fraction estimation and quality control against reference scRNA-seq data.