Ed Chen
Senior Product Designer · Design Lead · Institutional Finance
Senior PD, promoted January 2026 · Open to Principal PD / Design Lead / Head of Design · Global · Available Now
Five years shipping production design at a regulated ASIC broker — 100K+ traders, 40+ jurisdictions, $2B+ daily volume. A 150-component design system adopted across 5 product lines that absorbed 8 regulatory rewrites without a single rebuild cycle, with zero design-tied compliance findings in four years.
40+ regulated jurisdictions covered · zero material UX-tied compliance findings across four years · $2B+ daily volume on the platform the system ships to.
A senior / principal product designer who treats compliance as a product surface — shipped at a regulated broker, not theorised in a deck — and an AI-fluent designer who ships production code (Claude Code, Codex, Antigravity, Figma MCP). Everything below is supporting evidence.
- Senior / Principal Product Designer
- Design Lead
- Design System Lead
- Founding Design Engineer · AI-augmented
What the four-year record actually shows
Regulation read before Figma opens
Most designers need someone to summarize the regulation. Ed reads MiFID II Article 27, ASIC RG 268, and FCA COBS before the first wireframe. The design voice carries weight in compliance reviews because the preparation already happened.
Design systems that survive regulatory change
At ACY Securities, a 150-component system absorbed 8 regulatory rewrites without a rebuild cycle. Each event that used to cost 3–4 engineering weeks now costs 3–5 days. That compression shows up as gross margin the business keeps.
Institutional domain depth — not retail polish
FIX 4.4 order entry, TWAP/VWAP execution surfaces, GARCH volatility models, HMM regime classifiers, EDD onboarding under FinCEN and FATF. Not learned from articles — shipped in production at a regulated broker serving 100K+ traders with $2B+ daily volume.
Production velocity, with AI in the loop
Claude Code, Codex, Figma MCP, and Antigravity AI carry the marketing-asset and campaign load that used to consume design hours — 2–3 marketing/asset campaigns shipped per day in 2026, up from 1 per week in 2024. Trading-surface releases run on a slower cadence with full regulatory review; the AI tooling does not compress regulated work. The AI material on this portfolio is interactive demos, not slides.
Where Ed fits
Senior / Principal Product Designer
IC track at a trading platform, prime brokerage, or regulated fintech. Drives design direction end-to-end — owning stakeholder alignment across PM, engineering, legal, and compliance without a design manager translating in between. At ACY, this meant aligning 8 regulatory rewrites with 4 product lines across 40+ jurisdictions while the design system absorbed every change without structural rework.
Design Lead / Design Manager
For organisations growing from one designer to a small team, or building a design practice from scratch in a regulated context. 6 designers led, 4 hired personally across 4 years at ACY, with the leadership scope validated under regulatory pressure across eight rewrites and one solo year — proven leader at first-time-leader cost — promoted to Senior PD in January 2026, with the lead-tier scope ready to be formalised as a Design Lead role.
Design System Lead
Architect and govern a compliance-first component library at scale. Token-first architecture, multi-platform export (CSS / JSON / Tailwind), regulatory change absorption built in at the primitive level. Proven at 150 components across 5 platforms, with a working Storybook (edwson.github.io/Ed_Storybook) for engineering consumption.
Founding Design Engineer / Design Engineering Hybrid
For YC-stage or AI-native fintech founding teams. Ships working code — HTML/CSS/JS, React, TypeScript, Vite — alongside the design rationale. 8 concept applications have working React prototypes; Praxis Prime ships 5,288 LOC TypeScript as a deployable cPanel build. Bridges design and engineering without translation loss.
Four years of design leadership, documented in artefacts
Four pillars, each with its own evidence — cross-functional alignment, design ops, editorial restraint, and team-cycle management — built over four years of hiring, attrition, solo runs, and rebuild in a regulated context. The leadership-record disclosure sits below the pillars.
Cross-functional alignment as primary job
Aligned 5 product lines (Web · iOS · Android · Admin · Partner Portal) on a single component contract. Every regulatory rewrite went through a single weekly sync between PM (priority), Engineering (feasibility), Compliance (regulation interpretation), Legal (disclosure language), and me (design absorbing all four constraints into UI). The system never had a single design-tied audit finding across 8 rewrites and 40+ jurisdictions — that's the evidence the alignment process worked.
Design ops · system adoption mechanics
When a design system "ships" but isn't adopted by engineering, that's a design ops failure — not a component-library failure. The 150-component Edwson Design System reached 5-product-line adoption because the handoff process was designed deliberately: Figma library published quarterly, token export shipped as CSS/JSON/Tailwind so engineers could consume in their preferred format, Storybook (now v1.0 GA) as the source of truth for component behaviour, and a single Slack channel for design system change discussions. Design ops as a deliverable, not an afterthought.
Editorial restraint at the org level
The hardest part of design leadership in fintech isn't shipping — it's saying no to features that look great in isolation but violate the system's grammar. The Edwson Design System absorbed 8 regulatory rewrites without a rebuild cycle because the design org said no to one-off compliance treatments, no to feature-specific styles, no to bespoke regulatory disclosure layouts. Every change propagated through tokens, not new components. That discipline only exists when design has the authority to refuse.
Hiring, onboarding, and team-cycle management
Four designers hired personally — full interview loop, offer, onboarding, and a two-week ramp to first independent contribution via the self-service enablement system (Storybook + token reference + regulatory routing playbook). Led a team of four, ran solo through attrition, now rebuilding. The concept studies double as design-org curriculum — each carries an explicit "what I deliberately did not build" section that models the intake discipline I ask of every report. Full team-cycle record in the disclosure below.
Six designers led, mentored, or managed across four years — four hired personally through the full interview loop, offer, onboarding, and two-week ramp via the self-service enablement system (Storybook + token reference + regulatory routing playbook). The team moved through full headcount cycles: four designers, solo, four again, solo again, currently rebuilding. The newest direct report joined May 4, 2026 and is in Week 2 of the 2-week ramp as of writing — the enablement system is being empirically re-validated, not retrospectively described. No open requisitions at present.
The value proposition: a designer who has run design through hiring, attrition, sole-contributor mode, and rebuild — in a regulated context with eight regulatory rewrites absorbed without a rebuild cycle and zero material UX-tied compliance findings. For a Design Lead / Head of Design / Design Director role the next employer gets a proven leader at first-time-leader cost — capability validated under regulatory pressure, with the Senior PD title now officially earned and the lead-tier scope ready to be formalised one level up.
The title now matches the work: I was promoted to Senior Product Designer in January 2026, after four years already operating at design-lead tier. The scope was proven first and the title caught up — and the trajectory points one level higher, to Design Lead or Head of Design. Full record at career-journey.html — Chapter 2 covers the team-cycle history, Chapter 4 covers current state.
Shipped at an ASIC-regulated broker
Each case study has a named client, a regulatory anchor, and a measured outcome. Concept work is labeled separately and never conflated with shipped production.
150-component design system across 5 product surfaces. Absorbed 8 regulatory rewrites (MiFID II, ASIC RG 268, FCA COBS, FINRA) at the token layer — no surface-level rebuild. Covers 100K+ traders, 40+ jurisdictions, $2B+ daily volume.
- KYC drop-off: 73% → 45% at Enhanced Due Diligence
- Order execution time: 8.2s → 2.9s on institutional ticket
- Regulatory rewrite cost: 3–4 weeks → 3–5 days per event
- Zero material UX-related compliance findings across 4 years
Sole product owner, designer, and full-stack developer for the Christie's International Real Estate blog and community surfaces (christiesrealestate.com/blog). UHNW audience reading $5M–$80M luxury real estate; brand voice translated from a 260-year auction house heritage into a digital editorial register. Designed and built end-to-end alone — content templates, editorial system, social/community integrations, and the full front-end implementation.
Redesigned financial data widget platform. Controlled within-subjects usability study (n=15, paired t-test t(14)=8.92, p<0.001) on the order-placement task: 3.1s → 1.9s. Cohen’s d = 2.47 on the paired difference. Full methodology disclosure with seven named limits + eight-direction replication plan linked below.
Same token-driven system that ships at ACY, applied across seven fictional brands in seven adjacent regulatory regimes. Each study ships a working artefact, not a mockup — built from public regulatory frameworks (FINRA 2111, SEC Reg BI, FinCEN CDD, FATF Rec 10, BSA, SR 11-7, SEC Rule 15c6-2 T+1, CSDR), competitor teardowns, and domain-expert consultations. Zero client overlap, original IP.
- TradeX — Institutional terminal · FIX 4.4 · buy-side workflow research
- Xanthos — UHNW private bank · FINRA 2111 · SEC Reg BI fiduciary spine
- Argos — AML/CFT surveillance · FATF typologies · FinCEN SAR filing
- Aureus — B2C AI wealth hub · 4 AI modules · 10-step KYC + AML flow · Reg BI Decision Room
- ConnectX — Private-banking CRM · 8 named AI agents · 4 explicit autonomy tiers
- Double-Blind — Adversarial AI for UHNW relationship management · SEC 17a-4 · SR 11-7
- Praxis Prime — Front-to-Back prime brokerage · FIX 4.4 · SWIFT MT54x · T+1 (SEC 15c6-2) · CSDR · 5,288 LOC React/TypeScript
The institutional record proves regulatory depth; these prove 0-to-1 range. Each is designed, branded, and built solo and shipped to a live audience in the Japanese market — the same designed-honesty and fiduciary-alignment discipline, applied to consumer products.
- Duo-Shou — a LINE bot “second-hand strategist”: photo-to-valuation with calibrated AI confidence, sell-timing alerts, channel routing. No commission, no custody of money. Live in Japan.
- PawsRoam — a 0-to-1 B2B2C pet-services marketplace built from seven months of Tokyo field research to live MVP. Full product ownership.
Independent design & front-end · 2016–2020
Before and alongside the institutional work, a run of independent contracts — each one designed and shipped to production front-end, solo. Same through-line as everything else here: a designer who builds what he designs.
Landing pages & front-end builds — IBM, LA Auto Show, Sourcing at MAGIC, a GDC 2019 promo site (EVA), and independent brands. Designed and coded, 2016–2020.
Working surfaces, not slides
Every concept on this portfolio ships a working artefact. Click through, not just scroll through — these are the same primitives that live in the ACY production system, applied to six adjacent regulatory regimes.
Frameworks designed around in production
Where Ed's profile fits
Firms that operate under regulatory scrutiny and need a designer who treats compliance as a product surface, not an afterthought. Companies where design decisions have P&L and regulatory consequences. Available across US (East & West Coast), Europe (DACH · UK · Benelux), Asia-Pacific (Singapore · Tokyo · Sydney).
European wealth-management ecosystem note. Switzerland-based reviewers have been a consistent presence in 2026 traffic data. The Xanthos Private Bank + Aureus AI Wealth Hub concept studies are deliberately framed against the FINRA 2111 / SEC Reg BI / MiFID II / FINMA fiduciary spine — directly applicable to UBS-Pictet-Julius Baer / Lombard Odier / Mirabaud workflows. See Xanthos → · See Aureus →
What the work compounds into, in P&L terms
Design decisions in a regulated platform are not aesthetic — they sit upstream of conversion, compliance cost, engineering velocity, and audit exposure. The framing below maps four years of shipped work at ACY Securities to the financial line items a hiring committee actually owns. Where a number is modelled rather than measured, it is labelled as such.
Industry benchmark per ASIC enforcement action against an AFSL holder for retail-protection or disclosure breach, cited in project-acy.html against the published RG 78 + RG 234 guidance. Four years of zero material UX-tied compliance findings across 40+ jurisdictions translates to a defensible avoidance position; the exact count of "violations avoided" cannot be empirically established (you cannot count the thing that did not happen), but the cost-avoidance frame is the one a CFO will recognise.
Time-per-screen for the regulated trading-surface squad halved across Q3 2022 – Q4 2023. At a fully-loaded engineer cost of approximately $180K/year USD for a Sydney mid-senior FE engineer (Robert Half 2024 IT salary guide, public benchmark), the 50% velocity lift across a 4-engineer squad working ~180 screens/year is equivalent to ~$360K–$420K USD of engineering capacity returned per year, measured directly in Jira completion data not modelled. The mechanism: regulatory rewrites absorbed at the token layer instead of as per-screen rebuilds.
KYC funnel completion rate doubled after the 10-step disclosure-aware onboarding redesign, measured in GA4 across the Q2 → Q3 2022 window. At a conservative $400 LTV per onboarded retail trader (CFD industry benchmark, public broker disclosures) and ~1,800 incremental successful onboardings per quarter from the lift, the order of magnitude is ~$700K USD additional annual LTV from one design change. The exact LTV multiplier is a modelled industry figure, not ACY's internal LTV; the funnel-completion delta itself is GA4-measured.
MiFID II Art.27, ASIC RG 227 + RG 268, FCA COBS, FINRA 2111, FinCEN CDD, FATF Rec 10 + 12, OFAC sanctions propagation, and PDS/TMD revisions all landed without a system rebuild. Industry baseline for a regulated fintech of comparable size is one full rebuild cycle every 18–24 months at roughly 6–12 engineer-weeks per cycle; the avoided-rebuild figure across eight rewrites is order-of-magnitude ~48–96 engineer-weeks of avoided rebuild work. The token-first architecture is not free — it requires editorial restraint at the org level (Pillar 03 above) — but it makes regulatory change a configuration problem instead of an engineering problem.
How to read this. Each of the eight regulatory rewrites between 2022 and 2026 — MiFID II Art. 27, ASIC RG 227/268, FCA COBS, FINRA 2111, FinCEN CDD, FATF Rec 10/12, OFAC sanctions propagation, and PDS/TMD revisions — landed on the token layer, not on individual screens. Because disclosure logic, suitability gates, and severity states live as tokens and primitives, each change propagated as configuration across all five product surfaces instead of triggering a per-screen rebuild. The industry baseline for a regulated fintech of comparable size is one full rebuild every 18–24 months at roughly 6–12 engineer-weeks each; absorbing eight rewrites this way is order-of-magnitude ~48–96 engineer-weeks of avoided rebuild work. This diagram visualises the mechanism behind the “8 rewrites · 0 rebuilds” figure above — it adds no new claim. Source · ACY engineering retro records, 2022–2026
The honest aggregate. Adding these four lines together produces an order-of-magnitude figure on the scale of $1M–$2M USD per year of compounded engineering capacity returned, LTV captured, and rebuild work avoided — before counting the optionality value of zero ASIC findings. This is not a marketing claim; it is the math a CFO or CPO will run on their own back-of-envelope when evaluating a Senior PD or Design Lead hire. The senior-PD honesty register is that these are modelled aggregations of measured component metrics, not a single empirically-measured ROI number — the disclosure notes below name exactly what was measured and how.
How design actually compiles in 2026, when AI is in the loop
Most fintech designers in 2026 either (a) talk about AI without using it in production, or (b) use it everywhere without disclosing where. Neither passes the regulatory-context honesty test. The model below is what I actually run in 2026 — explicitly named tools, explicit scope, explicit limits. The same discipline that keeps regulated work safe applies to AI-augmented work.
What I delegate to AI
- Scaffolding and boilerplate — React component skeletons, Tailwind utility chains, JSON-LD schema drafts, TypeScript type declarations. Claude Code + Codex + Antigravity each carry the load when generation volume is the bottleneck.
- Marketing-asset and campaign load — the 2–3 polished assets per day cadence on hire.html refers to marketing-asset work specifically, not regulated trading-surface releases.
- Documentation drafts — case-study skeleton, JSON-LD FAQ candidate questions, alt-text first pass, i18n string batches across en/ja/zh-Hant lang packs.
- Code-base reconnaissance — "where does this token resolve", "show me every inline-style violation", "regenerate the calc-safe minified CSS". The kind of grep + audit work that used to consume an afternoon.
What stays under human judgment
- Regulatory interpretation. No AI suggestion ships into a regulated surface without the regulation being read first by me. The four disclosure notes below document exactly which regulations were read against which UI guardrails.
- The naming of the problem. AI is excellent at executing a well-named problem and dangerous at naming one. The first 30 minutes of any case study — what is the failure mode actually called — stays human.
- The disclosure register. Every "modelled, not measured" callout, every named limit, every honest negative result was written by me. AI tends toward enthusiasm; the disclosure voice is the opposite of enthusiasm.
- The sign-off. No AI commits to production without me reading the diff. The Praxis Prime case study explicitly discloses AI-augmented authorship of 5,288 LOC of React / TypeScript with human sign-off on every commit — that is the load-bearing pattern.
The Praxis Prime case study openly cites Claude Code, Codex, and Antigravity in the authorship trail. Hidden AI assistance reads as dishonest in a senior interview; disclosed AI assistance reads as a contemporary 2026 workflow. The portfolio chooses disclosure.
AI in this workflow is a multiplier on Ed's design judgment — not a substitute for it. The regulated trading-surface cadence runs on the slower, regulation-first rhythm. The marketing-asset and documentation cadence runs on the AI-augmented rhythm. The two cadences are deliberately different.
Every AI-augmented commit is reviewed and signed off at the diff before merge. The same SEC Rule 17a-4 / SR 11-7 effective-challenge discipline applied to AI in regulated finance applies to AI in this portfolio — the human stays accountable for the output, the tool stays accountable only for velocity.
Full disclosure of how AI was used on Praxis Prime · project-praxis-prime.html (5,288 LOC React / TypeScript, AI-augmented authorship via Claude Code + Codex + Antigravity, human sign-off at every commit). The four 2026-05-18 disclosure notes below extend the same register to regulatory methodology, post-launch iteration, accessibility, and enterprise-IT political skill.
Before forwarding this profile, read the four disclosure notes.
The shipped work and the seven concept studies sit on a deeper evidence layer: four field notes that disclose what is measured vs what isn’t — methodology, post-launch iteration, accessibility, and the enterprise-IT political-skill case. Each is summarised in the cards below.
The integrity claim is the disclosure register, not the headline numbers. Every quantitative claim on this page has either an externally citable source (ASIC Public Register, ACY institutional disclosures) or a named internal instrument (GA4, Hotjar, Jira, Mixpanel, Lookback.io). Where a number cannot survive scrutiny under the disclosure-note discipline, it is either labelled as modelled-not-measured or removed from this page. The four notes are how the engineering, compliance, and accessibility scepticism a hiring committee will bring is met head-on rather than left to chance.
If you are evaluating this profile for a Senior PD, Principal PD, Design Lead, or Head of Design role: the recommended 30-minute reading order is post-launch first (most applicable to a hiring decision), methodology second (the most statistically rigorous), then accessibility, then political-skill. The four notes together form the senior-PD honesty register the rest of the portfolio leans on.
- Finlogix Cohen’s d = 2.47 disclosure
- n = 15 paired within-subjects design
- Seven limits named
- Eight-direction replication plan
- RG 268 disclosure surface, 18 months
- v1 modal → v2 strip → v3 inline
- Muscle-memory dismissal failure named
- Honest negative result on AUM lift
- 12 token contrast pairs measured
- ARIA + keyboard patterns documented
- Six limits named
- Six-component replication plan with cost estimates
- ACY Connect, 12 institutional clients
- One IT director who didn’t want to migrate
- Three credible objections named
- Three verification surfaces that earned adoption
The regulation is already read. The domain is already deep. The system is already built. If your team needs a designer who can walk into a compliance review, a product sprint, and an engineering standup in the same day — and add signal to each one — let's talk.
Available for hiring conversations across time zones — US East Coast afternoon, EU morning, and Japan / APAC business hours (JST). Email is the fastest route and is actively monitored: ed@edwson.com · replies within one business day. Resume (PDF) and the full about page are linked below.