travr.se

Designing a cognitive search and smart documentation tool for software engineers

 

travr.se

Designing a cognitive search and smart documentation tool for software engineers

Travr.se was a tech startup that built a software tool to help software engineers document their code-writing process.

Goal: supporting engineers in complex code writing

We needed to understand where engineers felt least supported, and how our tool could help them over the barriers of documentation and communication. This meant measuring code productivity as well as thoughtfully exploring engineers' experience with tools.

Methods

Cognitive problem-solving observation

Behavioral user testing 

Semi-structured interviews and thematic coding

Cofounder & Chief Researcher

As research lead for travr.se, I led all parts of this project from prototype exploration, interviews, analysis, and presentation for investor pitches.

I was a founder of traver.se, with the wonderful Chap Snowden. Engineering work was also supported by the indomitable Kurt Collins. Both of these people are awesome.

My Contribution

Example Research Output

Over 55 semi-structured interviews with high tech engineers 

engineering team interviews 

3 pitch decks and research reports 

4 cycles of prototype designs 

Cognitive Search Maps for engineering workflow 

Example Research Question: 

Travr.se wanted to test how our features could support collaboration on engineering teams. To do this, we needed to understand engineers' problem-solving workflows and where communication broke down.

I used a mix of coded behavioral observation and cognitive mapping to generate "search maps" for our users. First, users allowed me to observe them solving a problem at work. Then, engineers sketched their own mental maps of problem-solving. I synthesized these and quantified similarities to help travr.se understand engineers' main pain points. Our product design focused on addressing these pain points. For example, via the travr.se tool we provided easier documentation layers that could be retrieved or hidden while browsing through shared code!

We built tool prototypes and tested them with dozens of users, following up in both interviews and surveys. 

 

One key design insight: engineers lost hours of time trying to find code examples and earlier problems. This led us to focus on a product that could help engineers easily discuss code examples and search through a "team library" of code draft work, even if that code never made it to production.

After several rounds of usability testing, we integrated text and comments on existing documentation, and built a web app for the travr.se product. 

One key design insight: engineers wanted to be able to leave comments or notes as they were researching online, and often thought documentation was woefully outdated and incomplete. Qualitative interviews revealed these hidden frustrations!

Google

Mixed-method research and data science to support learning and development

 

As part of the Metrics & Evaluation team in People Development at Google, I led research to support engineers and high tech talent.

Predictive analytics over large datasets 

Qualitative and semi-structured interviews 

Program evaluation, survey design and analysis 

Longitudinal learning outcomes 

Behavioral scenario testing

Methods

 

Design Lab

HCI Research on how software interfaces can frame feedback for online users

At the UC San Diego Design Lab, I was a postdoctoral fellow working with an interdisciplinary team of HCI researchers 

In collaboration with peerstudio.org, we wanted to test whether interface changes could change the way that online students gave each other feedback.

We also wanted to use the platform to help undergraduates provide each other with free essay and graduate application feedback!

Goal: empirically test whether interface design changes would impact online peer to peer learning.

Interface prototyping 

Quantitative statistical analysis 

User followup surveys and interviews

Methods

My Contribution: Senior Researcher

Research lead 

This was a very joint project on our team. I collaborated with Ailie Fraser, Vineet Pandey, and Scott Klemmer on these experiments, along with several research assistants in the lab. I supervised research design, which was collaborative. I conducted the statistical analysis for this project, and produced the majority of our writeups, but with significant input from my collaborators.

CHI paper, 2016: Framing Feedback

3 interface iterations to the peerstudio design 

Helped over 50 students with their graduate application essays, for free! 

Example Deliverables