
Ontologic · Research software
Making complex research workflows easier to follow
Turning researcher insights into a clearer path from choosing an analysis tool to reviewing its results.
- Industry
- Research technology / B2B SaaS
- Client
- Ontologic
- Role
- UX/UI designer
- Team
- Howard Kim, Darren Anorga, Bryan Deras
Archived project. Ontologic has shut down and its website is no longer active. This case study preserves the research and design work.
The problem: too much work around the work
Experimental researchers needed to organize data, choose analysis tools, and make sense of the outputs. Fragmented systems and inconsistent definitions made that process harder to follow, shifting attention from research to managing the workflow.
Our project focused on the moments where that uncertainty surfaced: entering the platform, selecting a script, and understanding what happened during a run. The design challenge was to provide clearer guidance while retaining the controls researchers needed.
My contribution
Working with Darren Anorga and Bryan Deras, I helped translate researcher interviews into journey maps, user flows, wireframes, and interactive prototypes. We used these artifacts to connect individual interface decisions to the larger research process.
How I framed the problem
I approached this as a sequence of decisions, rather than a collection of screens to refresh. Before changing the interface, I wanted to understand what a researcher needed to know in order to move forward—and where the product was asking them to make that connection on their own.
Find the right tool
Help researchers compare options using the context of their experiment.
Understand the next step
Make required inputs and the action to start a run easier to identify.
Keep the work traceable
Connect the inputs, analysis pipeline, and outputs across a run.
Understanding the researcher’s workflow
01 — Start with the people doing the work
We interviewed researchers who were using Ontologic to understand their needs and frustrations. A researcher persona brought the recurring needs together: organize scattered information, adapt data structures, and spend more time on insights.
What I took from the research
I used the persona to keep the researcher’s goal in view: getting to an insight. That helped me frame organization and data management as support for the work, rather than the end goal. The question I carried into the design was: what context does this person need here, and what is getting in the way?

02 — Look at the journey, not isolated screens
The journey map expanded the view from a single task to the full research process: collecting data, structuring it, making changes, and generating insights. It highlighted why continuity mattered—researchers needed to understand how one stage connected to the next.
Our user-flow work then focused on login confusion, uncertainty around script selection, and limited feedback during execution. These became priorities for a more explicit sequence of actions.
How I connected the findings
I read the journey as a series of handoffs. Choosing a script was only useful if the next screen made its requirements clear; running it was only useful if the output could be found again. This gave me a way to assess each screen by how well it prepared someone for the next step.

Designing a clearer path through analysis
We developed wireframes and interactive prototypes to turn the research into a more understandable workflow. The archived screens below show the design decisions at each stage.
These screens preserve different design iterations, including changes to the navigation. They document the project’s direction rather than a currently available product.
01 / Choose a tool
Make the options easier to compare
Script selection was a point of uncertainty. The tool library made comparison more concrete by pairing each tool’s name and version with a short description and research-specific tags.
Filters for experiment type, data type, and model gave researchers a way to narrow the list. “My Lab” and “Public” separated the scope of available tools without making people leave the library.

I wanted researchers to judge a tool before opening it. Names alone would not explain whether it suited their experiment, so descriptions, versions, and tags became part of the comparison. The tradeoff was a denser list, but the detail had a clear purpose: supporting a more informed choice.
02 / Prepare the analysis
Make the setup requirements visible
After choosing a tool, the question becomes what it needs to run. The Kilosort screen separates the recording data file from the channel map, names the supported upload formats, and groups the parameter fields below.
A visible tool version, compatibility information, and a GitHub reference provide context. “Run Kilosort” names the main action, while “Enter Sample Data” offers a separate way to explore the setup.

I did not want to hide the parameters researchers needed just to make the screen look simpler. Instead, I focused on grouping the requirements: files first, settings next, then the run action. Keeping the tool description and version nearby preserved context while the researcher prepared the analysis.
03 / Review a run
Keep inputs and outputs connected
The Run Log gives completed work a consistent structure. Each row connects the date and time to the input files, the pipeline, and the outputs. Search and lab/public filters help researchers find the relevant record.
Selection controls and duplicate, download, and share actions support follow-up work. The design makes a previous run something researchers can revisit, rather than an isolated result they have to reconstruct.

I treated the run history as part of the core workflow, not an administrative afterthought. A result needs context to be useful later. Putting inputs, pipeline, and outputs in the same row made that relationship visible, while search and filters provided a way back when the history became longer.
04 / Inspect the results
Let the analysis take center stage
The archived results exploration brings specialist visualizations into a single workspace. Cluster, waveform, feature, and trace views sit together, with upload, share, and download actions in the surrounding interface.
The important balance was clarity around the analysis without oversimplifying the analysis itself. Researchers still needed rich detail; the platform needed to make its location and available actions understandable.

My thinking · keep the expertise in the foreground
Here, I wanted the application to support the analysis without competing with it. The specialist views carry the meaning, so I kept the surrounding actions separate and recognizable. The design question was how to help researchers orient themselves while preserving the depth of information they came to inspect.
A consistent visual language
The project’s visual system used violet, pale surfaces, and Open Sans to connect the interface. Persistent navigation, repeated field treatments, and clear groupings gave the different workspaces a common structure.
The archived style references below show the palette and typography behind those screens.
How I used visual consistency
I viewed consistency as a way to reduce relearning between tasks. Repeated navigation, field treatments, and page groupings could give researchers familiar cues even when the content changed. The visual system’s job was to make the workflow feel connected, not to draw attention away from the data.
What the project delivered
The work translated researcher interviews into a concrete design direction: a more explicit route through tool discovery, setup, run history, and results. The archive preserves the research synthesis, visual foundations, and interface designs that made that direction reviewable.
Research translated into priorities
Interviews, a persona, and journey mapping connected the interface to researchers’ needs.
A workflow made visible
Wireframes and prototypes organized key decisions into clearer stages.
Context carried across tasks
Tool metadata, setup inputs, and run records made the relationships between steps easier to inspect.
Reflection
The central lesson was that a complex product does not become clearer simply by showing less information. It becomes clearer when the right context appears at the moment a person needs to decide.
For this project, that meant explaining the available tools before selection, naming the requirements before execution, and keeping inputs connected to outputs afterward.
What I would validate next
I would test whether researchers could select a suitable tool, prepare its inputs, and locate a previous output without assistance. Task completion, setup errors, time to find a run, and confidence at each step would help evaluate the proposed improvements.
The available archive documents the design work, but does not establish a verified post-launch measurement baseline.




