← Work
HR Tech

Harver

Designing for both sides of the hire: candidate-facing assessments grounded in neuroscience and behavioral science, and recruiter-facing tools that surface those insights as science-backed hiring decisions.

HR Tech Neuroscience Behavioral Science Machine Learning
My Role

I work across Harver and Pymetrics to solve complex product challenges in assessment experiences, platform integration and hiring decision support.

My focus is on making assessment journeys clearer for candidates, helping recruiters understand results with more confidence and shaping how two products come together into a more coherent experience.


Products I've worked on under Harver

Harver, formerly Outmatch, brought together a family of hiring and assessment products through multiple acquisitions. Across these products, I contributed to product migration, feature parity, and the candidate and recruiter experiences that helped bring the platform together.

Launchpad logo
Launchpad
LaunchPad Recruits was a London-based SaaS recruitment automation and video interviewing platform that helped employers screen candidates, run assessments, automate hiring workflows, support reviewer insights, and use predictive analytics to make hiring faster, more consistent, and less biased.
Outmatch logo
Outmatch
Outmatch was a Dallas-based hiring experience platform that combined pre-employment assessments, video interviewing, and reference checking to help companies predict candidate success and make faster, less biased hiring decisions. Built through a series of acquisitions including LaunchPad and Checkster, Outmatch acquired Harver in 2021 and rebranded under the Harver name.
Harver logo
Harver
Harver was an Amsterdam-born HR-tech company offering pre-employment assessment software for high-volume hiring. Its platform helped employers replace résumé-first screening with structured assessments, predictive matching, automation, and data-driven candidate selection, while improving candidate experience and reducing bias in the hiring process.
pymetrics logo
pymetrics
pymetrics was a New York-based AI hiring and talent-matching platform that used neuroscience-based games, behavioral data, and machine learning to assess candidates’ cognitive, social, and emotional traits, then match people to roles based on potential rather than résumés or traditional credentials.

01

Overview

Integration challenge

After Harver acquired multiple enterprise products, each product came with its own design system, user base, workflows, and operational complexity. The challenge was to unify them into one coherent platform experience while creating a seamless journey for both candidates and recruiters.

Design’s role

Product design helped shape the post-acquisition migration strategy through research, feature parity analysis, workflow mapping, and shipped product improvements. Instead of migrating every feature as-is, we evaluated what to carry over, adapt, simplify, or retire to create a more coherent platform experience.

Constraints

Live enterprise customers on both products, two engineering orgs, differing design languages, and a compressed integration timeline.


02

The Problem Space

User friction

Users of both platforms faced overlapping but inconsistent workflows — creating confusion about where to go, what to trust, and how things connected.

System fragmentation

Two separate design systems, component libraries, and interaction patterns made it impossible to ship consistently or efficiently.

Opportunity

The combined product had the potential to offer experiences neither platform could deliver independently — but only if integration was handled deliberately.


03

Design Workstreams

Platform Unification

Audited and reconciled interaction patterns, navigation models, and visual language across both products. Established shared design principles to guide integration decisions going forward.

Feature Parity

Mapped overlapping capabilities across both platforms. Defined which patterns to carry forward, merge, or deprecate — balancing user expectations on both sides with product direction.

New Product Initiatives

Identified and designed net-new experiences only possible through the combined product — surfaces and capabilities that created clear differentiated value post-integration.

Existing Product Improvement

Used integration as a forcing function to address UX debt, improve consistency, and raise quality across both products — turning a constraint into a design opportunity.


04

Process & Approach

Start with why

I started by understanding the integration challenge across Harver and Pymetrics through design audits, user research synthesis, product data, and conversations with users, stakeholders, and teams across both products.

Prototype the path forward

I translated research insights into shared design principles, user journey flows, and interactive prototypes that helped teams align on the right experience before committing to implementation.

Test & learn

I helped reconcile different user mental models, workflows, and feature expectations by testing ideas early, reviewing trade-offs with stakeholders, and defining what needed to migrate, evolve, or be left behind.

Ship together

I worked closely with product and engineering teams across both organizations to phase delivery, refine UI in the codebase, document decisions, and support a more unified platform experience.


05

Outcomes & Impact

User experience

Reduced cognitive load for users navigating across both platforms. Clearer workflows, less duplication, more intuitive wayfinding.

Design system

Unified component library and token system enabling faster, more consistent design and engineering output across both product surfaces.

Product direction

Net-new product surfaces shipped that were only possible through integration — demonstrating clear combined-platform value to customers.

Org alignment

Shared design principles and governance model established across two teams — creating a foundation for continued collaborative delivery.


06

Reflection

What I learned

Acquisition work forces clarity — you can't carry everything forward, so every decision becomes intentional. That constraint is actually good for design quality.

What was hard

Navigating two product cultures with different norms, tools, and assumptions about how design works — and finding shared language quickly enough to move.

What I'd do differently

Invest earlier in a shared research synthesis. The cost of misaligned mental models compounds fast when two user bases are both in scope.

Details abstracted to protect confidential product information.