Experience from a prior engineering role in collaboration with the University of Huddersfield & Crowther Accountants
Overview
The Audit Program was created to solve a familiar problem in accounting: audits were slow, manual, and heavily dependent on Excel spreadsheets. During a research engineering collaboration between the University of Huddersfield and Crowther Accountants, I contributed to the development of a system that automated the extraction, interpretation, and validation of financial data using intelligent automation techniques and the ChatGPT API.
The Challenge
Auditors often work with large, inconsistent Excel files filled with unstructured data. The manual process of:
- Extracting relevant fields
- Checking for discrepancies
- Validating entries
- Identifying anomalies
…was slow, error‑prone, and mentally draining.
The project needed to solve three core problems:
- Interpret messy, unstructured Excel data automatically
- Detect anomalies without relying on traditional ML models
- Present everything in a clean, accountant‑friendly interface
All while respecting the sensitivity of financial data and maintaining compliance.
What We Built
The solution combined a modular backend with an interface designed directly with accountants, ensuring the system felt familiar rather than disruptive.
Key Capabilities
- Automated extraction of relevant data from complex Excel sheets
- Intelligent interpretation using the ChatGPT API
- Automated discrepancy detection
- Secure data handling and storage
- A simple, approachable UI tailored for accountants
Technical Approach
- Backend: Python + Django
- Database: MySQL
- Automation: ChatGPT API for natural‑language‑driven data interpretation
- Architecture: Modular, extensible, and easy to maintain
- UI: Designed collaboratively with accounting teams
One of the most interesting aspects was using ChatGPT to interpret data structures that would normally require custom ML models. This reduced development overhead while increasing flexibility.
Impact
The Audit Program delivered meaningful improvements:
- Up to 40% reduction in audit processing time
- Automated validation and anomaly detection
- Fewer manual errors
- More time for accountants to focus on analysis rather than data wrangling
- A smoother, more predictable audit workflow
It demonstrated how modern automation can elevate traditional accounting practices without replacing the human expertise behind them.
Why This Experience Matters
Although this project predates Cosmokode Ltd, it reflects the same principles we bring to our work today:
- Use automation to remove friction
- Design tools that feel intuitive, not intimidating
- Build systems that adapt to real‑world workflows
- Leverage modern AI where it genuinely adds value
This experience directly informs how Cosmokode Ltd approaches intelligent automation and workflow design.
Tech Stack
- Python
- Django
- MySQL
- ChatGPT API
- HTML/CSS/JS for UI
This case study reflects prior professional experience and is included to demonstrate capability. Cosmokode Ltd does not claim ownership of the Audit Program.
