Tower.dev
End-to-end design ownership across product, branding, and front-end implementation for a data infrastructure platform built to run AI-powered data apps at scale.
Services
product design
Branding
Front-end
Team
7 People
Table of contents
(02)
—
(03)
—
(04)
—
(05)
—
When I joined Tower the product existed as code and conviction. My job was to turn it into something developers would trust on first sight. I started by running a founding team workshop that went deep — users, goals, success metrics, brand direction, and competitive positioning all mapped before a single frame was opened. From that session I built a research and design blueprint that structured the entire engagement: competitor analysis, Kano modelling from user interviews to separate table-stakes features from genuine differentiators, feature prioritization, and a full information architecture. The product design covered the flows that matter most for a developer tool — a frictionless onboarding experience, the critical path from adding your first data app to running it in production, environment and settings systems, and the full spectrum of system states. I then designed and implemented the marketing website, giving Tower a public face that could hold its own in a competitive infrastructure market.
When I joined Tower the product existed as code and conviction. My job was to turn it into something developers would trust on first sight. I started by running a founding team workshop that went deep — users, goals, success metrics, brand direction, and competitive positioning all mapped before a single frame was opened. From that session I built a research and design blueprint that structured the entire engagement: competitor analysis, Kano modelling from user interviews to separate table-stakes features from genuine differentiators, feature prioritization, and a full information architecture. The product design covered the flows that matter most for a developer tool — a frictionless onboarding experience, the critical path from adding your first data app to running it in production, environment and settings systems, and the full spectrum of system states. I then designed and implemented the marketing website, giving Tower a public face that could hold its own in a competitive infrastructure market.
From Pipeline to Production: Live
02
UX Research
UX Research
Every product decision I made at Tower had a research reason behind it. Over 20 interviews with founders, engineers, and CEOs across data and AI companies gave me the raw material — a detailed picture of how technical teams operate infrastructure today, where they lose time, and what they had stopped expecting tools to solve for them. A Kano model analysis structured that material into a clear product direction, shaping which features made it into the core experience and which defined Tower's edge in the market. A competitor audit across the leading orchestration platforms confirmed where the gaps were and where design could do something the category had not done before. The information architecture, the onboarding flow, the environment system, the monitoring layer — each one was designed with a specific research finding in mind.
Every product decision I made at Tower had a research reason behind it. Over 20 interviews with founders, engineers, and CEOs across data and AI companies gave me the raw material — a detailed picture of how technical teams operate infrastructure today, where they lose time, and what they had stopped expecting tools to solve for them. A Kano model analysis structured that material into a clear product direction, shaping which features made it into the core experience and which defined Tower's edge in the market. A competitor audit across the leading orchestration platforms confirmed where the gaps were and where design could do something the category had not done before. The information architecture, the onboarding flow, the environment system, the monitoring layer — each one was designed with a specific research finding in mind.
03
Design System
I built the design system through working sessions with the development team where we aligned on which shadcn/ui components we would use, how we would name and structure our shared semantic language, and how design decisions would map directly to code. Beyond the token layer, I set up the full design system infrastructure: component documentation with usage guidelines and do/don't examples, a versioning system to track changes across design and code in sync, contribution guidelines so the system could scale without breaking consistency, and a changelog maintained throughout the project so developers always knew what had changed and why. Grounding the system in shadcn/ui meant the development team never encountered an unfamiliar token name in the handoff — every component I delivered was immediately implementable, reducing overall time to production and giving us full dark mode support without any rework.
I built the design system through working sessions with the development team where we aligned on which shadcn/ui components we would use, how we would name and structure our shared semantic language, and how design decisions would map directly to code. Beyond the token layer, I set up the full design system infrastructure: component documentation with usage guidelines and do/don't examples, a versioning system to track changes across design and code in sync, contribution guidelines so the system could scale without breaking consistency, and a changelog maintained throughout the project so developers always knew what had changed and why. Grounding the system in shadcn/ui meant the development team never encountered an unfamiliar token name in the handoff — every component I delivered was immediately implementable, reducing overall time to production and giving us full dark mode support without any rework.
04
Final Designs
Every design decision in this product traces back to a single research finding: founders do not need more data, they need to know what their data means. That insight shaped everything from information hierarchy to how I approached both the happy path and edge cases. Financial products rarely break on missing data — they break on conflicting data, and I designed for exactly that: KPIs contradicting each other, metrics sitting outside plottable benchmark ranges, and forecast-to-actual divergence significant enough to distort chart scale. The drilldown panel was the most technically complex component, built to move a founder from a headline number to a full diagnosis in three clicks rather than three tools. The benchmark visualization required the most iteration — a raw percentage told founders nothing, so I replaced it with a segmented positioning bar that communicates peer standing instantly without calculation. Light and dark mode were built in from day one through the semantic token system, so no component ever needed to be designed twice.
Every design decision in this product traces back to a single research finding: founders do not need more data, they need to know what their data means. That insight shaped everything from information hierarchy to how I approached both the happy path and edge cases. Financial products rarely break on missing data — they break on conflicting data, and I designed for exactly that: KPIs contradicting each other, metrics sitting outside plottable benchmark ranges, and forecast-to-actual divergence significant enough to distort chart scale. The drilldown panel was the most technically complex component, built to move a founder from a headline number to a full diagnosis in three clicks rather than three tools. The benchmark visualization required the most iteration — a raw percentage told founders nothing, so I replaced it with a segmented positioning bar that communicates peer standing instantly without calculation. Light and dark mode were built in from day one through the semantic token system, so no component ever needed to be designed twice.
05
Conclusion
Building this product end to end, from the first founder workshop through to a shipped token-driven design system, reinforced something I already believed but now understand more deeply: the best financial tools do not show you data, they help you think. The user interviews and mental models that shaped the architecture never left the room — they stayed present in every component decision, every status label, and every annotation in the developer handoff. The most valuable skill I developed here was designing for trust. Every component, every edge case, every piece of the system was in service of a founder who needed to make a real decision with real consequences. The product is now entering its Seed round, which is the best possible signal that the design did its job.
Building this product end to end, from the first founder workshop through to a shipped token-driven design system, reinforced something I already believed but now understand more deeply: the best financial tools do not show you data, they help you think. The user interviews and mental models that shaped the architecture never left the room — they stayed present in every component decision, every status label, and every annotation in the developer handoff. The most valuable skill I developed here was designing for trust. Every component, every edge case, every piece of the system was in service of a founder who needed to make a real decision with real consequences. The product is now entering its Seed round, which is the best possible signal that the design did its job.
More works
Lifedata.ai
A platform for Omnichannel Customer Experience.
Lifedata.ai
A platform for Omnichannel Customer Experience.
Tower.dev
A Python-native infrastructure platform.
Tower.dev
A Python-native infrastructure platform.
Contact me
I Am All Ears!
Contact me
I Am All Ears!
Contact me
I Am All Ears!
The kind of a designer that feels like a keeper.
About
Contact
© 2014-2025. Designed by zdoor ruslan.
The kind of a designer that feels like a keeper.
About
Contact
© 2014-2025. Designed by zdoor ruslan.
The kind of a designer that feels like a keeper.
About
Contact
© 2014-2025. Designed by zdoor ruslan.