PathwayDNA

  • Traditional career quizzes are rigid, one-dimensional, and rely on dense corporate jargon that overwhelms younger users while failing to give adults actionable economic data.

  • Created an adaptive UI system inspired by architectural neo-pop aesthetics and teen-accessible UX principles. The interface dynamically morphs based on age mode—surfacing hand-drawn future personas and low-stakes micro-quests for teens, or executive gap matrices, salary floors, and 4-pathway ROI models for adults

AI-Accelerated Product Development Approach

"Leveraging Autonomous AI Agents to Compress 6 Months of Clinical Research, Generative Prototyping, and Production Engineering into Days."

Rather than using AI as a surface-level novelty or simple copy generator, I used an Agentic design-engineering workflow. AI acted as a multi-specialist pair programmer: validating psychometric scoring rubrics, generating targeted aesthetic assets, translating live research benchmarks, and continuously testing UI code in an isolated sandbox.

My role As the Product Lead and Design Engineer, I conceptualized, directed, and built PathwayDNA from o0 to 1. I steered the end-to-end product strategy, authored the multi-model psychometric architecture, established the adaptive neo-editorial visual system, and directed AI agent workflows to move from blank slate to a fully functional, cloud-ready production platform.

Key features

"Future You" Teen Character Spotlight
(Dashboard in Student Mode)

What I Led & Owned

1. Product Strategy & Vision (The "Why" & "What")

  • Identified the Market Gap: Recognized that existing career tools are either dry corporate assessments or superficial quizzes that force users down traditional degree paths while ignoring trades, apprenticeships, and burnout risks.

  • Defined the Core Value Proposition: Conceived the concept of a unified "Career DNA"—fusing 5 clinical models (MBTI, Big Five, Enneagram, DISC, RIASEC) with behavioral swipe telemetry and public earnings data.

  • Dual-Audience Product Framing: Architected the dual-experience framework that serves both teenagers (zero jargon, aspirational future-self personas) and adult career-switchers (salary floors, 10-year net ROI, and skills-gap matrices).

2. Creative Direction & Visual Systems (The Aesthetic & UX)

  • Iterative Design Direction: Rejected generic corporate SaaS templates and dusty palettes; steered the branding through rigorous visual iterations toward a modern, high-contrast Ungrammary-inspired neo-editorial system (38px architectural grid, Acid Volt #D0FC2C, Electric Magenta #EE03FF, and crisp typography).

  • Persona Art Direction: Defined the visual brief for the hand-drawn, graphic-novel style character personas (Maya, Devon, Dr. Samira, Kai), ensuring they looked aspirational, mature, and relatable rather than childish.

  • UX Micro-Interactions: Designed key interactive patterns including the Tinder-style swipe deck that morphs the live Holland Hexagon, the 7×67 \times 67×6 Trait-to-Skill Compatibility Heatmap, and the 3D capsule progress capsules.

3. AI Agent Orchestration & Prompt Engineering (The Process)

  • Directed Autonomous AI Subagents: Acted as the orchestrator of multiple specialized AI agents (psychometric authoring, image generation, backend engineering, testing), defining strict constraints and validation criteria.

  • Ensured Clinical & Mathematical Rigor: Supervised the authoring of 194 reverse-keyed test items, ensuring psychometric validity, bias control, and mathematically sound trait-vector matching algorithms.

  • Quality Assurance & Verification: Established rigorous automated testing criteria inside sandboxed environments to verify SQLite persistence, multi-user isolation, and zero-error API responses.

4. Design-Engineering Execution (The Build)

  • Full-Stack Architecture: Supervised the implementation of an asynchronous Python 3.11 Tornado server and modular ES6 frontend with zero heavy framework bloat.

  • Production & Cloud Readiness: Configured infrastructure-as-code deployment (render.yaml) with persistent disk storage to prepare the platform for real-world user testing and iteration.

MVP Published in record time

I used Antigravity agents to iterate on Information Architecture and Layout in real time.

The Visual identity inspired by Ungrammary's neo-editorial drafting grid and high-contrast pill architecture."

Teen visual future-self engine

midjourney iteration cycles for a quick mvp, not a final product

"Solving teen career anxiety: replacing sterile charts with hand-drawn, relatable future-self personas."

The quality of the graphics, colors and overall look and feel need a series of visual design Charettes and iterations, and perhaps a few sessions of usability tests with teens. Next steps would be to invest in visual designers who can own the end-to-end visual aesthetic of all the unique characters and persona’s we’ll land on.

Next project

View Case Study

Or continue reading about what I learned on this project below. Keep scrolling!

What I learned

Defining what’s viable

Balancing design progress with collaboration across global teams is crucial.

  • Defining guardrails around viable solutions is necessary, even when user feedback is prioritized.

  • Web-based tools can provide real-time data access and analysis that excels has limitations with.

  • Clear and consistent terminology is essential for user understanding.

Complexity of Enterprise UX: I am not the user.

Working in the Seller Space meant that every UX Challenge had 3 underlying dimensions: an Operational Process improvement challenge, A physical vs Virtual status challenge, and an information display or UI Challenge. This is quite a complex and yet fascinating space to work. On top of that, dealing with legacy tools where separate programs add features in silos and an evolving user base meant having to tackle complexity to arrive at an optimum UI solution. The risk is that there are many obstacles to making intuitive design changes that would benefit the Seller Base in very significant ways. 

Development Resource Constraints vs. User needs


Excel has its limitations
Excel has functionality in analyzing and transforming data that cannot be duplicated in a custom web tool. A dashboard provides simplicity of access at a glance, allowing users to quickly gain insight into key metrics without the complexity of having to download a file and perform analysis. Additionally, a dashboard provides information that may not easily fit into an excel spreadsheet, such as item images or trending customer comment themes. Dashboard stays up to date with new data in real time while eliminating the seller workflow of having to download and re-analyze new data from a new files. This is particularly important for sellers who have a significant number of returns but have not yet established a streamlined process around handling them. 


Email cannot replace a web-based tool
Emails don’t provide access to data in real time and require the seller to establish a manual workflow around receiving them, logging back into seller central and conducting an investigation into single return orders. For sellers who handle multiple return orders per day, the current emails received do not allow sellers to gain a full understanding of the return data. 

Identifying Quick wins for UX
Quick wins identified include 1) Making the terminology in reports clear & consistent by a) Provide tool-tip on hover states with clarifying explanations of business-centric terminology that may be unfamiliar to sellers. b) Revising terms in column titles in fulfillment reports such as unfulfillable, unsellable, sellable, customer-damaged, defective & damaged.