3. Data, Science & AI

Generates and validates Schema.org-compliant JSON-LD for SEO-rich snippets, including automatic page detection and Google Rich Results testing.

Generates interactive reports with multi-format output and data visualization capabilities.

Provides RAG-specific utilities for score interpretation, query optimization, and source naming in document search and ingestion workflows.

Translates natural language user input into executable system commands through intent recognition and NLP processing.

AI-powered engine that proactively recommends relevant actions based on contextual user behavior and interaction patterns.

Enables end-to-end implementation of machine learning pipelines, model serving, and feature engineering for production deployment.

Builds LLM applications, RAG systems, and prompt pipelines for AI-powered features, chatbots, and intelligent applications.

Enables data analysis, SQL query execution, and BigQuery operations to derive actionable business insights.

Optimizes prompts for large language models and AI systems to enhance performance in AI feature development and agent interactions.

Enables efficient SQL query writing, execution plan optimization, and database schema design for data management and performance.

Builds ETL pipelines, data warehouses, and streaming architectures for scalable data processing and analytics infrastructure.

Optimizes database performance through query tuning, schema design, and migration management for efficient data operations.

Manages behavior log entries in a local database, supporting recording of behavior events, Pavlok API responses, and coach comments.

Configures ReinforceNow training runs using config.yml and train.jsonl, and converts HuggingFace datasets to ReinforceNow format.

Provides command-line interface for Reinforcement Learning from Human Feedback (RLHF) training, enabling project setup, training runs, and model management.

Enables writing reward functions for Reinforcement Learning training in ReinforceNow, including @reward decorators and LLM-based evaluation.

Formats training data in JSONL format for ReinforceNow, handling entry structures, tools, rewards, and sandbox setup.

Generates training data from teacher models for off-policy distillation, enabling efficient knowledge transfer in AI model training.

Performs quality control analysis for pooled CRISPR screen data, assessing library representation, read distribution, and gene recovery to ensure reliable hit calling.

Automates metagenomics analysis from FASTQ data to taxonomic and functional profiles using Kraken2, Bracken, and HUMAnN.

Processes and annotates differential gene expression results from DESeq2 and edgeR, including significance testing and gene annotation for downstream analysis.

Aligns short DNA reads using Bowtie2 with flexible modes for genomics applications like ChIP-seq and ATAC-seq.

Specialized tool for lipid identification, quantification, and pathway interpretation in LC-MS lipidomics using LipidSearch, MS-DIAL, and LipidMaps.

Creates and manipulates bigWig tracks for genomic data visualization, converting bedGraph to bigWig and extracting signal values using UCSC tools and pyBigWig.