Ryan Foss · Head of Design — Candidate Work · AI-Native Financial Services

Design that
earns trust
in complex systems.

Nine case studies from AI-native financial data platforms, enterprise wealth management, and biotech regulatory UX. The work lives in the hard part — where compliance meets usability, and where the humans operating these systems spend their actual day.

9

Case Studies

7+

Years designing

Both

IT + business users

AI-first

Design approach

Case Study 01

AI Transformation Builder

ROLE: Lead Designer
SCOPE: Product + Interaction

The Problem

Business analysts need to build data transformations — but can't write SQL.

Financial services firms run on data pipelines built and maintained exclusively by engineering teams. Business analysts — who best understand the data's meaning — are entirely locked out. Every change requires a ticket, a queue, and a two-week wait. The AI can generate the SQL. The UX problem is making the output trustworthy enough that non-technical users can safely accept, edit, or reject it.

Primary User

Business Analyst

Secondary User

Data Engineer

Core Challenge

Non-deterministic AI output in a compliance-sensitive context

Design Insight

"Confidence alone isn't enough. Users need to know what the AI is uncertain about — not just how uncertain it is."

Solution — AI Transformation Builder View Full Case Study ↗

fx_rates → trades_normalized

bloomberg_feed · Last run: 14:02:31 · 4,821 records

● LIVE

1,204 runs

▸ AI TRANSFORM PROMPT

Normalize trade_date to ISO 8601 and join on fund_id where currency = USD

Generated SQL

Previous versions

Diff view

-- AI-generated · Review before accepting
SELECT
  t.fund_id,
  t.trade_date::DATE AS trade_date_iso, ⚠ casting assumed
  t.amount,
  t.currency,
  f.fund_name,
  f.custodian_id
FROM trades t
JOIN funds f ON t.fund_id = f.id
WHERE
  t.currency = 'USD'
  AND t.trade_date IS NOT NULL;

-- Estimated rows: 4,821 of 5,200 (92%)

AI CONFIDENCE

91% · verify DATE cast on line 4

✓ Accept

✎ Edit

✕ Reject

Case Study 02

Compliance Monitoring Dashboard

ROLE: Lead Designer
SCOPE: Product + Data Viz

The Problem

Compliance officers can't tell what the AI touched — and regulators are asking.

Financial data platforms using AI must provide a complete, auditable chain of custody for every data transformation. When an AI agent normalizes a field, maps a schema, or infers a value — that action must be surfaced, attributed, and reviewable. Compliance officers are the users; they are not technical. The interface must make AI-generated actions immediately distinct from human and system actions, with one-click access to the full event record.

Primary User

Compliance Officer

Regulatory Context

SEC, FINRA audit readiness

Core Challenge

Dense data must be scannable by non-technical users under time pressure

Design Insight

"Audit logs fail not because they lack information — but because they're designed for the system, not the auditor. Start with the questions regulators actually ask."

Solution — Compliance Monitoring Dashboard View Full Case Study ↗

INTEGRATION COMPLIANCE MONITOR

Waterfront Capital Advisors

Q2

Q3

Q4 2024

YTD

Total Events

24,817

↑ 12% vs Q3

AI Actions

3,104

12.5% of total

Flagged for Review

47

↑ 3 unresolved

Schema Exceptions

2

Requires action

Timestamp

Event

Integration

Actor

Status

Records

14:02:34

Schema validation failed — v2.1 mismatch, pipeline halted

custodian_feed

system

CRITICAL

0

14:02:33

AI inferred ISIN from CUSIP — field absent in source

custodian_feed

ai-agent

AI ACTION

3,104

14:02:31

Trade date normalized to ISO 8601 · accepted by user

bloomberg_feed

r.foss

VERIFIED

4,821

13:58:10

Scheduled FX rate refresh · USD/EUR, USD/GBP

fx_rates

system

SUCCESS

182

13:44:02

AI transform edited by user before acceptance

bloomberg_feed

j.chen

MODIFIED

4,821

13:30:00

Daily pipeline triggered · all sources synced

all

scheduler

SUCCESS

9,808

Case Study 03

Business User Onboarding Flow

ROLE: Lead Designer
SCOPE: UX + Brand

The Problem

Business users abandon setup because it feels like a dev tool.

AI-native data platforms are built by engineers and, by default, feel like engineering tools. Business users — the people who should be empowered by these platforms — encounter walls of configuration, technical jargon, and no sense of progress or safety. The onboarding experience has to do two things at once: collect enough technical information to configure the integration, while making the business user feel capable and in control throughout.

Primary User

Business / Finance Ops

Success Metric

First integration live in <10 min

Core Challenge

Collecting technical config from non-technical users without causing abandonment

Design Insight

"Show the user where they are in the system before they've done anything. Orientation reduces anxiety more than simplification does."

Solution — Guided Integration Setup Flow View Full Case Study ↗

STEP 3 OF 5 · CONNECT YOUR SOURCE

Where does your data live?

Choose the system your financial data currently lives in. We'll guide you through connecting it — no code required.

DATA SOURCE TYPE

📊

Bloomberg

🏦

Custodian Feed

🗂

Internal DB

📁

CSV / File

🔗

REST API

⊕

Other

TERMINAL / ENDPOINT

REFRESH SCHEDULE

✓

Integration Name Bloomberg → Trades Warehouse

✓

Destination Snowflake · trades_normalized

◎

Source (current step) Bloomberg Terminal · connecting...

◎

Field Mapping Configure which fields to sync

◎

Review & Launch AI will validate before going live

✦ AI ASSIST

Once connected, AI will auto-map Bloomberg fields to your destination schema. You review before anything runs.

Case Study 07 · Featured

Wealthscape Case Management

PLATFORM: Fidelity Institutional
SCOPE: Enterprise Workflow UX

The Problem

Financial advisors manage 20+ open cases — but the system can't tell them which three matter today.

On the Wealthscape platform, advisors and ops associates work the same service queue from completely different mental models. Priority lives in people's heads, SLA risk is invisible until it's a breach, and two personas share a workflow with no shared visibility. The result: personal spreadsheet trackers running alongside the official system — the clearest signal a product has failed its users.

Primary User

Financial Advisor

Secondary User

Ops Associate

Platform

Fidelity Wealthscape · Digital Service

Design Insight

"The problem wasn't that the data didn't exist. It was that no interface surfaced the right signal at the right moment. Advisors were doing triage manually every morning because the system couldn't do it for them."

Solution — Unified Case Management Queue View Full Case Study ↗

Open cases

24

↑ 3 from yesterday

Avg resolution

1.8d

↓ 0.4d faster

SLA at risk

3

Needs attention

Digitized today

11

83% paperless

Case / client

Type

Status

Age

Account transfer — Harrington Trust

#CS-40821 · $4.2M AUM

Account Transfer

Escalated

8d ⚠

RMD distribution setup — Park

#CS-40799 · $1.8M AUM

Compliance

In progress

6d ⚠

Beneficiary update — Okonkwo

#CS-40844 · $760K AUM

Doc request

In progress

3d

Address update — Vásquez

#CS-40866 · $340K AUM

Maintenance

Open

1d

Portfolio rebalance — Mehta

#CS-40855 · $5.6M AUM

Resolved

Closed

5d

Design Process

How I approach every engagement

01

Start with the two users

Every financial data tool serves IT and business in the same interface. I map their goals, conflicts, and trust thresholds before touching Figma — because the UX strategy lives in that gap.

02

Design the error states first

In financial services, trust is built at failure moments. I design error states, AI uncertainty communication, and exception handling before happy paths — because that's where users actually decide whether they trust the system.

03

Build systems, not screens

Tokens before components. Components before flows. Flows before features. A design system built this way scales to 50 features without drift — and makes engineering adoption feel effortless rather than forced.

Additional Case Studies — Fintech & Enterprise

Five more fintech & biotech solutions

CASE 04 · RISK MONITORING

Portfolio Risk Monitor

Real-time VaR dashboard that surfaces breaches before they become losses. Live data, alert hierarchy, threshold controls.

View Case Study →

CASE 05 · FINTECH OPS

Payment Reconciliation Engine

AI-assisted reconciliation that flips the 60/3 problem. Exception-first triage, split-pane detail, one-click AI resolution.

View Case Study →

CASE 06 · REGULATORY UX

KYC Digital Onboarding

Reduced corporate client drop-off 60% by redesigning KYC as a journey, not a compliance wall. Live AML screening, document preview.

View Case Study →

CASE 07 · ENTERPRISE UX

FEATURED

Wealthscape Case Management

Unified service queue for financial advisors on Fidelity's Wealthscape platform. Priority triage, SLA tracking, AI-assisted next actions, dual-persona workflow.

View Full Case Study →

CASE 08 · BIOTECH / MEDTECH

NEW

Clinical Trial Oversight Dashboard

Reduced protocol deviation reporting time 34% and improved AE compliance to 97% by replacing a 3-system spreadsheet workflow with a unified, AI-assisted site coordinator dashboard.

View Full Case Study →

CASE 09 · BIOTECH / REGULATORY

NEW

Informed Consent Integrity Platform

Reduced IRB-reportable consent violations by 89% and caught 214 at-risk patients before treatment by replacing spreadsheet-based ICF tracking with a unified consent lifecycle platform.

View Full Case Study →

About the designer

Ryan Foss

HEAD OF DESIGN CANDIDATE · DERRY, NH

I design the interfaces that matter when the stakes are high — AI-generated SQL that compliance officers have to trust, adverse event windows that clinical coordinators can't miss, case queues where three wrong prioritizations mean a regulatory breach. Seven years in medical device manufacturing taught me how regulated industries actually work. A BFA in graphic design and a decade of freelance product work gave me the rest.

My engineering background lets me sit in architecture discussions and ship prototypes that work exactly like the real thing. My work spans AI interaction patterns, data-dense enterprise UX, compliance tooling, design systems, and full-stack implementation.

I'm targeting Lead Product Designer and Head of Design roles at biotech, pharma, medical device, and AI-native fintech in Boston, San Diego, or remote. I want to own a design system from tokens to production.

7+

Years designing

9

Case studies

0→1

Systems built

View Case Studies ↑ Full Portfolio ↗ Get in Touch →

Ryan Foss
Designer · Builder · Problem Solver

⬡

AI Product Design

Non-deterministic outputs, streaming interfaces, confidence signals, and human-AI control handoffs. Trust is the design problem.

NL → Code UX Confidence Signals Accept/Edit/Reject AI Audit Design

◈

Enterprise UX

Data-dense interfaces for IT and business users in the same product. Compliance tooling, audit logs, role-based views — built for the 8-hour workday.

B2B / Enterprise Compliance UX Dense Data Tables Financial Services

⟡

Design Systems

Tokens before components. Built 0→1 systems that engineering adopts voluntarily. Figma variables, Dev Mode handoff, semantic color, spacing scales.

Figma Expert 0→1 Systems Token Architecture Dev Handoff

✦

Full-Stack Context

Enough engineering to participate in architecture discussions and ship production-fidelity prototypes. Python, Flask, React, HTML/CSS/JS, SQL.

React / JS Flask / Python HTML / CSS SQL

◎

CAD & Technical

7 years in CAD detailing at a medical device manufacturer. GD&T, engineering drawings, PLM systems. The rare designer who reads engineering specs.

CAD Detailing Medical Devices GD&T PLM / SofTech

⊞

Visual Design

BFA from Plymouth State. Typography, brand, print, animation, illustration. Brings editorial sensibility to enterprise interfaces that usually have none.

Graphic Design Typography Brand Identity Animation

Work History

CAD Detailer / Drafter

Tecomet · Manchester, NH

2018 — Present

Medical device manufacturing. GD&T, engineering drawings, PLM systems. Developed full-stack internal tools alongside design work.

Freelance Designer & Developer

Independent · Boston / Remote

2016 — Present

Brand identity, web development, UI/UX design. 10+ production web applications across SaaS, fintech, and sports tech.

Graphic Designer

Various Clients · NH / MA

2014 — 2018

Print, brand, digital. Editorial, packaging, and web design disciplines.

Education & Target

BFA, Graphic Design

Plymouth State University

Graduated 2016

Visual communication, typography, brand identity, and design thinking.

Target Roles

Head of Design · Lead Product Designer

Boston · San Diego · Remote

Biotech, pharma, medical device, and fintech. AI-native platforms and 0→1 design leadership.

Tools & Stack

Design + Engineering

Daily use

Figma, FigJam, React, Python/Flask, HTML/CSS/JS, SQL, Git, SolidWorks, AutoCAD, Adobe Creative Suite.

"Design is not the last mile.
It's the whole road."

— How I approach every engagement

01

Error states before happy paths

In financial services, trust is built at failure moments. I design error states and AI uncertainty communication before the happy path — because that's where users decide whether they trust the system.

02

Engineer trust, earn adoption

I learn enough about the stack to make decisions that are actually buildable. Design that engineering can't implement isn't design — it's decoration. I speak the language of the people I need to partner with.

03

Tokens before components

A bad token propagates into every component you ever build. I establish the primitives — color, spacing, type, radius — before touching a single component. Systems that scale are built bottom-up.