Deep OriginDrug discovery AI assistant2023 · 2024Lead Product Designer

Balto · an AI assistant for drug discovery.

At Deep Origin, I designed and shipped features across the Balto AI assistant and the core scientific platform. My work focused on helping scientists analyze complex molecular datasets, speed up research workflows, and make faster, more confident decisions in the drug discovery process.

Role

Lead Product Designer · Balto & platform

Team

Cross-functional product + science org

Category

SaaS · Scientific platform

Focus

AI assistant · Data viz · Workflow

Balto · login screen with molecular artwork
01 · Problem

Scientists were drowning in specialized tools.

Drug discovery scientists rely on multiple specialized software tools for modeling, data analysis, visualization, and comparison. Each platform has its own interface, workflows, and learning curve, forcing researchers to spend significant time switching contexts and mastering different systems instead of focusing on scientific insight.

This fragmentation slows down decision-making, increases cognitive load, and delays critical research milestones · at the exact moments where speed and confidence matter most.

"I spend more time learning software than I do interpreting results."

A recurring theme across every scientist we interviewed.

Drug discovery workflow diagram showing fragmented tools across research stages
Drug discovery workflow diagram showing fragmented tools across research stages
02 · Solution

One unified interface · analysis, visualization, comparison.

Balto integrates analysis, visualization, and comparison into a single unified interface. By connecting data sources and automating complex workflows, the tool enables scientists to query results, compare compounds, and generate insights through natural language and intelligent recommendations.

Uploads became informational and metrics-oriented rather than opaque imports. A team dashboard surfaced tasks, shareable knowledge, and success metrics · so the AI output lived inside a shared scientific workflow, not a chat sidebar.

The result: scientists can focus on interpreting results rather than navigating software.

01
Embed
Worked alongside scientists through real research cycles.
02
Unify
Reduced fragmented tools into one unified interface.
03
Augment
Layered AI for comparison, querying, and recommendations.
04
Measure
Tracked time-to-insight and decision confidence.
Balto · Hello Dr. Joe team dashboard
Balto · Hello Dr. Joe team dashboard
Screening results · 150 compounds, A2A Adenosine Receptor pipeline
Screening results · 150 compounds, A2A Adenosine Receptor pipeline
Caffeine physicochemical property analysis with 3D and 2D structures
Caffeine physicochemical property analysis with 3D and 2D structures
"
We went from juggling six tools to doing the work in one. Balto made it possible to ask a question and actually get an answer.
· Research scientist · Deep Origin
−50%
Reduction in time spent analyzing compound data
+64%
Faster decisions with AI-driven comparative analysis
$31.7M
ARPA-H contract award (up to) · in-silicon models
1
Unified interface, replacing many
03 · Reflection

AI in expert tools is a second pair of hands, not a second brain.

Designing for scientists taught me that AI in a domain tool has to be legible, auditable, and always in service of the human decision. The assistant's best feature isn't answering questions · it's showing its work in a way that an expert can challenge, trust, and build on.

Thanks to Balto, Deep Origin is now leading an up-to-$31.7M ARPA-H contract award to build advanced in-silicon models designed to help replace animal testing · accelerating safer and more ethical drug development.

Next case · 03 / 04
Co-founder · 2014–20