• 

• 

• 

• 

• 

• 

Inspiration

Syndicated loan operations still rely heavily on manual checks using PDFs, emails, and Excel. During our research, we found that Loan Agents often spend hours rechecking interest calculations before sending notices to multiple lenders. Even small errors can lead to payment breaks, follow-up notices, reputational damage, and audit issues.

We were inspired to build SyndiClean to address this last-minute risk. The idea was simple: if banks already follow standard formulas, why not automate the final validation step before notices are sent?

What it does

SyndiClean is a pre-send validation tool for syndicated loan interest notices.

It allows Loan Agents to upload an Interest Payment Notice (PDF) and automatically:

Extract key financial data (principal, interest rate, dates, interest amount)

Recalculate the expected interest using standard banking conventions

Clearly indicate whether the notice passes or fails validation

Explain any difference in plain, business-friendly language

The tool helps catch errors before lenders are notified and before money moves.

How we built it

We built SyndiClean as a lightweight web application focused on clarity and reliability:

Python for business logic

Streamlit for a clean, banker-friendly interface

PDF text extraction to read loan notices

Rule-based validation logic that mirrors how Loan Agents check calculations today

At the core is the standard interest calculation used in syndicated loans:

Interest

Principal × Rate × Days 360 Interest= 360 Principal×Rate×Days ​

The application presents results using clear green/red indicators and actionable explanations, without technical jargon.

Challenges we ran into

Inconsistent PDF formats: Loan notices are not standardized, making data extraction challenging.

Balancing strictness and usability: The system must reject unclear notices while still being helpful to users.

Designing for non-technical users: The interface needed to feel trustworthy and familiar to banking professionals, not engineers.

We intentionally limited scope to one high-risk use case to ensure reliability and clarity.

Accomplishments that we're proud of

Built a bank-grade MVP that solves a real operational pain point

Reduced a manual, error-prone process to a single-click validation

Created a tool that feels like a realistic internal bank pilot

Delivered a clear demo that shows time saved and risk reduced

What we learned

In banking, risk reduction matters more than flashy technology

The most valuable fintech solutions are often the most “boring”

Clear explanations build more trust than complex models

Focusing on one critical control can deliver immediate business value

What's next for SyndiClean

Future enhancements could include:

Support for additional notice types (drawdowns, fees, amendments)

Covenant calculation and breach alerts

Multi-currency and different day-count conventions

Integration with loan servicing platforms for straight-through processing

Our long-term vision is for SyndiClean to become a standard validation layer in syndicated loan operations, quietly preventing costly errors before they reach the market.

Built With

• git
• pandas
• python
• streamlit