Data Platform & Analytics Engineering

Gives agents a controlled Postgres inspection path for schema, table, and query context without turning database access into ad hoc shell work.

Lets data teams query business context across connectors when analytics questions span tools rather than one warehouse table.

Adds SQL Server execution-plan review to the workflow for query tuning and database performance investigations.

Keeps SQL and dbt analysis deterministic under coding agents, which matters when agents review warehouse changes.

Connects Airflow and warehouse operations through agent-safe data engineering workflows instead of unbounded DAG manipulation.

Covers Supabase/Postgres query, index, and RLS tuning where app data teams need database advice tied to production constraints.

Checks dbt model and warehouse parity before migrations or refactors are trusted.

Translates and validates SQL across dialects so warehouse migrations and multi-engine analytics work stay reviewable.

Catches risky PostgreSQL migration patterns before irreversible schema mistakes land.

Provides interactive inspection of large CSV files before teams map, clean, or ingest messy tabular data.

Helps analysts profile and triage messy tabular files quickly before building downstream transforms.

Adds plan-and-preview discipline for warehouse SQL model changes before rollout.

Connects agents to Airflow DAG context and orchestration workflows for platform-owned pipelines.

Lets agents inspect and operate dbt Cloud context without treating the warehouse as a black box.

Frames dbt model runs and transformation orchestration as a repeatable analytics engineering workflow.

Extracts model lineage so teams can reason about downstream impact before changing data assets.

Adds pre-commit gates for dbt projects before style, tests, or docs drift.

Lints and formats SQL so warehouse changes are reviewable and consistent.

Provides a fast SQL lint/format path for teams that need lightweight query quality checks.

Runs local analytics over structured files for fast investigation without provisioning a warehouse.

Queries CSV, JSON, and YAML exports with SQL before loading them into a larger system.

Turns many APIs and files into SQL-queryable surfaces for agents and data operators.

Creates explicit data quality expectations and validation artifacts for pipelines.

Inventories live schemas and produces reviewable docs before risky SQL or database changes.

Adds a local vector-indexing layer for retrieval and RAG experiments before teams adopt heavier managed vector infrastructure.

Grounds agent-assisted data discovery in catalog metadata, ownership, schema, and lineage context.

Gives agents bounded, schema-aware ClickHouse analytics access without handing them ad hoc database credentials.

Provides production hybrid retrieval and agentic RAG APIs with citations and reviewable document operations.

Adds document ingestion and search quality tuning before AI apps or agents consume retrieval context.

Converts PDFs, tables, images, audio, video, and web pages into structured text/markdown so RAG pipelines receive reviewable multi-format inputs.

Adds a mixed-document normalization step for analytics and RAG ingestion pipelines that need Markdown, JSON, CSV, or HTML before downstream processing.

Stores embeddings beside application data in Postgres for semantic search, RAG, recommendations, and agent memory retrieval.

Helps teams choose, evaluate, and rerank retrieval embeddings before feeding context into RAG workflows.

Adds a scalable vector database path for filtered similarity search across RAG and agent retrieval workloads.

Assembles and evaluates enterprise RAG workflows across internal data, models, and business users.

Gives analytics teams a portable context layer for packaging large reference sets without standing up a full service.

Adds a graph retrieval option for analytics teams working with relationship-heavy corpora and entity context.

Lets agents inspect governed Neo4j graph data through MCP instead of copying graph context into prompts.