3. Data, Science & AI

Enforces consistent data naming, directory structures, and metadata requirements for lab data, customizable to specific lab conventions.

Enforces customizable quality control thresholds for genomic data analysis, ensuring consistent pass/fail criteria during read filtering and alignment assessment.

Provides streaming primitives for bioinformatics file parsing and analysis without external CLI tools, supporting BAM, FASTQ, VCF, and genomics operations.

Efficiently extracts structured data from websites and APIs using programmatic scraping techniques for ETL pipelines and data processing.

Tracks data lineage and maintains audit trails to ensure reproducibility and traceability of data processing and analysis workflows.

Enables genomic ML workflows with transformer models (DNABERT, Nucleotide Transformer), quality-aware tokenization, and HuggingFace integration for sequence data.

Provides a real-time psychological dashboard for assessing energy, focus, emotions, cognitive biases, and burnout risk.

Provides domain-specific patterns for metagenomics analysis, including taxonomic profiling, diversity metrics, and streaming processing of 16S/shotgun metagenome data.

Automates quality control and sequencing statistics workflows in bioinformatics, generating comprehensive QC reports and assessing data quality metrics with full primitive coverage.

Monitors input context complexity in LLM applications and triggers state dumps to maintain model quality before degradation occurs.

Enables efficient cloud data streaming from S3, GCS, and HTTP sources without local downloads, supporting constant-memory analysis.

Provides standardized, validated pipeline templates for laboratory analysis workflows, enabling quick setup of new analyses with pre-approved structures.

Enforces strict bioinformatics project constraints: using only primitives, avoiding external tool calls, and ensuring deterministic, typed analysis.

Provides efficient batch processing patterns for handling multiple data samples, including parallel execution and memory-optimized result aggregation.

Builds production-ready machine learning and deep learning models using PyTorch or Scikit-Learn with mandatory ONNX export, following industry best practices.

Applies mathematical modeling, numerical analysis, and computational mathematics techniques as taught at top universities like MIT and Oxford.

Manages persistent workflow state using DuckDB for analytics, enabling dependency graph queries, historical metrics, and checkpoint recovery from markdown files.

Manages Supabase database schema, Row Level Security policies, and Edge Functions for secure and efficient backend logic.

Provides core architectural patterns for local AI agents, including Reflection, Planning, Tool Use, Multi-Agent, and Memory.

Expert review of hydrological models, statistical methods, and machine learning applications for SAPPHIRE forecast accuracy and scientific validity.

Manages knowledge graph connections between journal, task, and habit entities using neural clustering to derive AI-driven insights.

Demonstrates an AI-powered linter that validates code quality using large language models to detect issues and suggest improvements.

Extracts text content from PDF documents using PyPDF2 for data processing tasks.

Captures insights from errors to update agent context and prevent recurrence in future interactions.