All case studies
AI & Automation£52KSoftware Development

CODES AI (Internal)

AI Development Framework

0Dev Time Reduction
Timeline: Ongoing (started Nov 2025)Team: 2 engineers

Summary

Pioneered an AI-assisted development framework using local LLMs to automate data warehouse design, standardize ETL pipelines, and manage Azure cloud infrastructure code.

The Challenge

Traditional data engineering projects require weeks of repetitive boilerplate — schema design, ETL pipeline code, infrastructure templates, and testing. Senior engineers spend 60%+ of their time on code that follows predictable patterns rather than solving novel problems.

Our Solution

Built a comprehensive AI-assisted development framework using local LLMs (Claude, custom models) that automates the repetitive parts of data engineering. The system generates data warehouse schemas from business requirements, creates ETL pipeline code from data mappings, and produces Azure infrastructure templates from architecture diagrams.

Architecture Flow

Prompt Layer

CRAFT framework promptsContext-aware templatesProject-specific rulesCode style guides

AI Processing

Claude Opus reasoningCursor agent generationKimi 2M context refactoringAutomated testing

Output Layer

Production PySpark codeAzure IaC templatesCI/CD pipelinesDocumentation

Our Process

1

Requirements in English

Engineer describes what they need in plain English — 'Build a star schema for sales data with SCD Type 2 on customer dimension'.

Natural LanguageCRAFT Framework
2

AI Architecture Generation

Claude analyses the requirement and generates the full database schema, ETL mapping document, and architecture diagram.

Claude APICustom Prompts
3

Code Generation

Cursor agent mode generates production PySpark code, SQL DDL, Data Factory pipelines, and Azure DevOps YAML.

CursorPySparkSQL
4

Automated Review

AI reviews generated code for security (OWASP), performance (query optimization), and best practices compliance.

Claude Code ReviewSonarQube
5

Deploy & Monitor

One-click deployment to Azure with automated testing, monitoring dashboards, and performance baselines.

Azure DevOps CI/CDDatabricks

Results & Impact

0
Faster Development

Reduced development time across all data engineering projects

0
Code Output

Engineers produce 3× more deliverables per sprint

0
First-Pass Quality

AI-generated code passes review with minimal changes

0
Security Issues

Automated OWASP audit on every generated module

Technology Stack

Claude (Anthropic)Local LLMsPythonPySparkAzure DevOpsDatabricksDelta LakeCursor IDEFastAPI

Start a similar project

Let's discuss how we can deliver the same results for your business. Free consultation, no commitment.

More Case Studies