Building the NYE design org: nine hires, one design language, three platforms in fourteen months

At a glance

User

Product, engineering, and compliance partners shipping NYE across four squads — and everyday money consumers on the resulting app

Problem

No design function; late rework; four visual and verbal languages inside one super app

My mandate

Build the team and give NYE one experience before launch

Hard decision

Build the function first, then the system, then the product — and let the team overrule the plan where the plan was wrong

Result

Nine hires, one design language, −30% design lead time, three platforms in fourteen months

People affected

Hiring began in my third month and ran at one to two people a month until the team was skills-complete by month nine — seven full-time hires and two contract, who together could take a flow from research to motion without waiting on a vendor. I hired for the skills the team lacked rather than the strongest portfolio in the pile: a researcher and a UX writer came in before a third product designer. Product design: Shubham Sachdeva (Sr.), Aditi Arya, and a senior product designer. Research: Richa Deo (Sr. UX Researcher). Visual & motion: Abhilash Jose (Lead), Shrineh Namdeo (Visual Designer II), a senior motion designer (contract). Content: Subalakshmi Murugesan (UX Writer), Saumya Rai (Copywriter). Marketing: a copywriter and a marketing visual designer, also reporting to me. I also took on marketing. Acquisition creative, brand and in-app experience were pulling in different directions; putting a copywriter and a marketing visual designer under the same roof as product design meant the story a customer saw in an ad was the story they met in onboarding.

The apparent problem

Nobody owned the design pipeline end to end, and late-stage rework was pushing timelines.

The problem underneath

Each squad had its own patterns, so the brand eroded a little with every release — four products drifting into four visual and verbal languages.

My mandate

I owned

Build the design function and give NYE one experience before launch — team, operating model, and design system as the alignment mechanism.

Others owned

Product owned roadmap prioritisation per squad. Engineering owned delivery capacity. Compliance owned regulatory gates.

Final decisions

Product and engineering leads made release calls; the design team owned critique, system, and research centrally.

Team

11 led · 9 hired (7 full-time, 2 contract) across design, research, content, motion, and marketing.

Authority

Senior Manager UX — built and led the function; designers embedded per squad with shared critique, system, and research.

Delivery constraints

Fintech launching under regulatory pressure; four squads shipping in parallel; no pause for a complete design system before first release.

Constraints

• No existing design function — work scattered across squads and vendors.
• Late revisions eating the calendar before launch.
• Four product languages eroding the brand with every release.
• Regulatory pressure on disclosure-heavy flows such as savings-account KYC.

Critical decision

How the system changed

1. 01

   Critique — Twice a week, more during the launch push. The designer names the decision they want challenged; we critique the decision, not the designer.

2. 02

   Review & sign-off — Weekly, cross-functional: product, engineering, compliance. A flow leaves the room approved or with a named owner for the fix. No silent rework.

3. 03

   Standups — Designers sit in their squad's standup. We killed the separate design standup within weeks; it duplicated information and delayed it.

4. 04

   1:1s — Weekly with each designer; fortnightly with engineering and product leads, to surface conflicts before they became escalations.

5. 05

   Research — Rolling, not per-feature: continuous studies in Maze and moderated sessions, with findings in a shared Confluence repository every squad could read.

6. 06

   Stack: Figma as the single source of truth, Maze for research, Miro for workshops, Jira and Confluence for the pipeline. The point wasn't the tools; it was that there was one of each and everyone used it.

7. 07

   The NYE design system was the answer to "four products, one app": tokens, a component library, pattern guidelines and a content style guide, built so that wallet, UPI, banking onboarding and investments shared one grammar for onboarding progress, trust cues and money decisions.

8. 08

   Compliance disclosures as design patterns — in-flow components the compliance team approved once. Savings-account KYC onboarding went through compliance review first time because every disclosure it used was already an approved pattern.

9. 09

   Five principles anchored customer-facing decisions: convenience, transparency, trust, personalisation, support. The product decisions themselves are in the NYE product case study (/work/nye).

Validation and iteration

• Fully centralised → embedded

  I'd planned a central pool taking requests. Squads couldn't plan around it, so we assigned a named designer to each squad while keeping critique, system and research central. Ownership went up; queue time went down.

• Separate design standup → squad standups

  Dropped. The separate design standup duplicated information and delayed it.

• Full design system up front → tokens and core components first

  I'd scoped a complete system before the first release. The team argued for shipping a small token set and the components the first flows actually needed, then growing it. They were right — the full system would have arrived after launch.

• Per-feature research → rolling research

  Feature-by-feature studies couldn't keep pace with four squads. Continuous research with a shared repository could.

Outcome

Scale figures describe the operating context. Personal contributions are stated separately.

Frames

What I would change now

Resistance to a centralised team is about ownership, not design. I won squad leads over by giving them a named designer and a seat in review, not by arguing about org charts. It's great to be proven wrong early. Four of my structural decisions were overturned in the first quarter, and the operating model that shipped the product was the one the team built. Evaluate people by inputs, not output. In a launch year, output is hostage to engineering capacity; judgement and craft aren't. Delayed communication is worse than delayed action. Most of the churn we removed was information arriving late, not work done badly. What I'd do differently: publish a living pattern inventory with squad ownership from month one, and count retired duplicate flows, so the organisational evidence is as visible as the product.