Startup Product Design


Financial Analytics Tool


150 hours so far



UX Design Lead


November 2019

- Present 




Adobe Illustrator


The US municipal bond industry is an archaic industry that has been slower to adopt technology than other financial sectors. A fundamental problem with the industry is that municipalities file unstructured and non-standardized financial statements. This makes analyzing the underlying fundamentals of a municipality a costly and time intensive endeavor. 

We are building a centralized platform with standardized information and AI-enabled predictions for credit rating changes, which will help streamlines the municipal credit evaluation process.

$4 Trillion

Municipal bond industry

in US


Municipal bond analysts


300 Hours

Spent on analyzing bonds per year per analyst



I'm the solo UX Designer on the award-winning startup team.


I work with an experienced, diverse team that consists of three business professionals, a machine learning expert and a data science expert. Komodo won 1st place in the CMU annual Hack-a-Startup Contest and 3rd place in Princeton TigerLaunch Venture Pitch Competition. We are looking to participate in other startup competitions nationwide as well.

I (center) present to contest judges at CMU Hack-a-Startup


Dashboard that forecasts future credit changes to help municipal bonds analysts monitor credit risks


Home Dashboard.jpg
Part 1.gif


Aggregate information about each credit factor to provide a consolidated view of a municipality's health and outlook.


Downloadable data sheets of standardized financial data to streamline the process and avoid human errors. 

Part 2.gif


Municipal analysts spend hours collecting data and are forced to navigate to multiple data providers to evaluate a single bond.

In order to evaluate a bond, an analyst will need to research on four major credit factors: financial performance, economic strength, environmental, social and governance (ECG), and debt profile. To have a good understanding of all four areas, an analyst spends hours searching and pulling from different sources such as US Census and services like Merritt and Bloomberg.

Analysts also rely heavily on the Comprehensive Annual Financial Report (CAFR) that each municipality publishes. However, a fundamental problem within the industry is that municipalities file unstructured and non-standardized CAFR's in PDF formats, which makes extracting data difficult and analyzing the underlying fundamentals of a municipality costly and time intensive.

Desktop - 1 (5).png


Our team talked with over 70 analysts to understand their day-to-day work flow analyzing municipal bonds.


Municipal analysts look at multiple data sources and spend an average of 30 minutes compiling information per bond, with much of this time spent scanning and copying data from PDF files. As shown in the journey map, as analysts head to more and more sources to gather the data they need, their experience tank.

Artboard (3).png


Each analyst evaluates 12 bonds per week, which sum up to a total of 300 hours spent on just gathering data.


So, how might we help municipal bond analysts more efficiently evaluate credit risks?

We summarized the key user needs collected from municipal analysts, and brainstormed a few key features for our product. Based on these, I built the early wireframes for user testing.

Key User Needs

Decrease the amount of time spent on gathering municipal data.

Avoid human errors caused by manually copying and pasting data from PDFs

Make money by selling bonds before they drop in value and buying bonds before they increase in value.

Proposed Solution Features

Provide aggregated and standardized data to give municipal bond professionals a consolidated view of a municipality.

Allow analysts to instantly download information or export data directly into excel.

Building predictive analytics to provide real-time credit evaluations. (Long-term goal)


I followed an agile process by constantly user testing design concepts with municipal analysts and iterating.

One of the biggest challenges of this project is to acquire the domain knowledge needed to build an efficient service for municipal bond analysts. To overcome this barrier, I try to conduct as much user testing as possible with real industry professionals and iterate based on their feedback.


Low-fidelity wireframes

Frame 1 (11).png

“I love how concise this (design) is. A lot of the times you get on different website and things become messy… this is nice.”


      — Patrick Strollo, Sr. Municipal Analyst

I (left) conduct user testing using wireframes with Mr. Strollo (right).


Designing the credit evaluation system

A major feature of Komodo is its predictive analytics, which provides insights into a bond's performance in the future. I had to iterate over the design of evaluation systems several times to make sure that it both benefits our users and reflects our algorithm capabilities.

Users first said the rating systems were confusing.

The percentage values in front of "hold" and "sell" were confusing and lack of context.

More information was needed to gauge the influence of each credit factors beyond just relative increase and decrease.

So I started exploring different scoring systems with a few considerations in mind. 

Once decided on the percentile system, I started testing the score card layout

Artboard (4).png

Comparing two presentations of percentile, visual-heavy design is more direct and can deliver information faster.

A five-point scale can be effective in indicating the current performance of each credit factor.

Group 71.png

Information too clustered

Group 70.png

Having multi-colored bar is confusing and de-emphasize a bond's actual performance.

Group 41.png

Blue highlights can be mistaken as hyperlinks

Improved design incorporating a percentile system and a five-point system


Highlighting credit risks 

Analysts repeatedly bring up that they wish Komodo can help them identify which bonds need to be analyzed first based on recent changes and credit risks. I iterated over the home dashboard design to optimize this feature.

Initial design highlights all new updates and does not bring emphasize to those that are at risks

Took out some color indicators to red flag credits at risk

Get rid of unnecessary 

visual hierarchy and add in clear alert symbol



Home Dashboard 

Through the home dashboard, analyst can have a straightforward of how his account is doing overall. Since an analyst's most important job is to make sure he gets rid of risky bonds before they default, we highlight those potentially risky bonds and notify new updates that has caused fluctuations, so the analyst knows which bonds to prioritize in the evaluation process.

Home Dashboard.jpg




Pain point solved: Analysts often don't know which bonds to evaluate first.

Instead of showing the whole portfolio, we focus on showing bonds whose performance have decreased, since these are the bonds that need to be evaluated the most.


Pain point solved: Currently, there is no way to be notified when a municipality publishes new document.


We provide ​updates when municipalities publish CAFR, press release, meeting minutes and other documents, so analysts don't have to constantly look them up.


Portfolio Management, Watchlist and Advanced Screening

In addition to dashboard, we also included features to manage portfolios, curate watchlist and use advance screening to find products with customized filters. ​

Dashboard > Portfolio > Screening.gif


Bond Overview: Everything you need to know about a bond.

Every bond in the system will have a bond overview page, which provides high-level information about a bond and serves as the homepage to all data sheets associated with the bond.

Part 1.gif

Score Cards

Pain point solved: Only a portion of all bonds are rated by professional rating agencies.


Analysts are used to refer to ratings to understand the value of a bond, therefore those bonds that are unrated are less popular and less credible, despite being good bonds. We want to provide a credit analysis system that's comprehensive and unambiguous.


Therefore, with the users' need and the technical capabilities in mind, I designed the percentile rating systems, where a bond is compared to other bonds by state, sector and professional rating (if it has one). Then, by combining all three and calculate, we generate an overall percentile score, which appears in summaries and lists shown in home dashboard and portfolio.

Credit Factors

Group 124.png

Pain point solved: Information about municipal bond is fragmented, unstandardized and scattered.


We collect all information about a municipal bond and categorize them into five different factors. Then, we give each factor a rating on a 5-point and use the ratings to inform a bond's position in terms of percentile.   

Group 127.png

We want to be transparent with how we calculate our score and give users the power to customize.


Different factors are weighted differently, and we should the distribution under the "Score Facts" tab. However, we also know that an analyst might have a different view of how the score should be distributed, so we give them the option to adjust the weight of each factor. 

Group 126.png

Pain point solved: Analysts have to manually track historical performance of a bond.


Lastly, we provide a history of the bond's performance and an analyst can adjust the settings to see different views.


Financial Performance: From unstandardized PDFs to organized data sheets.  

Pain point solved: Analysts are reading 100+ pages PDFs issued by each municipality and copying/pasting data points manually, making the process error-prone and inefficient.


Financial Performance page of each bond provides data that are extracted from the CAFR document, standardized and put in to a clean data sheet format. In addition to existing data, we also provide our algorithmic prediction of the financials for the upcoming year. Analysts can easily export the sheet to Excel and other format and directly insert them into their reports. 

Part 2.gif


Economic Strength: Census data collection and visualization

Economic Strength page of each municipal bond will contain collected census information about the service area of each municipality. Each card has a visualization and a data table view that user can toggle between, and user will have the option to download as well.

Economic Strength.gif
Part 3.gif


Style Guide

Blue color scheme that is accessible both in light and dark


On top of a spectrum of blue, I picked red, yellow and green as color signals, since they already speak the financial language, to avoid reinventing the wheels. 

Color Scheme in Dark Mode

Home Dashboard.png
Economic Strength.png

I designed the logo to convey Komodo's brand image — to revolutionize something archaic 


I also leveraged Google Material Design for fonts and icons in this project to ensure maximum accessibility.

I built out cards, lists and other components with simplicity in mind


Analysts already face information overload. Therefore, in my design, I want analysts to consume information in the most simple and clear way possible.



We estimate that our solution can save analyst around 4 hours per week, which saves a company around $2000 per week. 


The project is an on-going one. We are user testing with municipal bond analysts every week to collect feedback on the current design. Additionally, we are also working with potential investors to create mock-ups that better showcase our vision.

I will continue to iterate on the high-fidelity prototype and validate user's needs through conducting user feedback sessions. As we continue to participate in venture capital competitions, I hope to get feedback from industry experts as well. All in all, this has been a challenging yet fruitful experience, and I hope we will become successful in bring Komodo to life.

CMU McGinnis Venture Capital Competition Pitch Video

I produced and edited this video