Problem
Handshake hosts a vast number of student profiles, accessible through multiple touch points on the employer platform. However, the way these profiles are displayed varies inconsistently, with 2-3 different layouts depending on where employers access them. This inconsistency creates confusion for recruiters, making it difficult to scan and compare candidate information efficiently—especially when reviewing long lists of potential hires. Additionally, the student-created profiles do not visually align with what employers see, leading to a disconnect between the two sides of the marketplace. 

Opportunity
Given the inconsistencies in profile displays and the difficulty of parsing key candidate information, our primary goal was to ensure that recruiters and employers see the most relevant details upfront, wherever candidate information appears. This required reevaluating not just the profile detail page but also list views, feeds, and profile preview cards. We saw a significant opportunity to enhance both the structure of candidate data and the overall browsing experience across match lists, search results, and event sign-up pages.

Process
I led the design efforts on the employer side, collaborating closely with a design partners on design systems and the student product. Given the new profile structure, I initially designed some of the key components myself and worked with the design systems team to determine where we could introduce new elements or reuse existing ones. My design partner provided valuable insights into the student profile structure, helping us align the two experiences and create a more cohesive system. Throughout the project, I worked closely with my PM—from drafting the PRD and analyzing research to preparing deliverables for engineering.

Research
My product partner and I first analyzed the existing research, benefiting from multiple rounds of interviews our UXR team had already conducted with recruiters and SMB owners. These interviews provided valuable insights into the information they prioritize and the challenges they face. Additionally, we conducted a small sample study, asking recruiters to rank profile information based on importance. The findings helped us determine the hierarchy of content within the highlights module and preview cards, ensuring we surfaced the most relevant details effectively.

Outcome
In Q4 2024, we launched the first version of our updated candidate profile page, leading to a 20% increase in message sends within the matches experience. This was a major win for the team, and the success of the design paved the way for the next round of iterations, building on our initial vision to enhance the list view and browsing experience in the candidate feed.

Problem

Blind has always been a platform for tech workers to talk about work, company information research, and finding new opportunities in the industry. However Blind as it stands is limited to research alone, with no way to take any action on companies or open roles that they see being talked about on the platform. Many job seekers end up asking for referrals or open positions through organic posts, which most of the time do not lead to any help or results.

Opportunity

We realized that there is a very clear opportunity here to connect those who are researching and looking for jobs directly with the most popular companies and their open roles on Blind. Furthermore, by creating a job board on Blind, these jobs can further fuel the conversations between our users when they are comparing / contrasting offers and pay ranges with each other.

Our primary measurement of success on the consumer side is job views / sessions per week, and on the b2b side, the amount of jobs we can pull in from either existing job boards or new roles directly from internal recruiters.

Process

I managed the design teams work across the b2b and b2c side of jobs, where I personally worked on building out the consumer side of the experience, and had 2 designers on my team building out the b2b jobs portal. The consumer experience was focused on browsing, filtering, and searching for open roles, while the b2b side was focused on job post creation as well as importing existing roles from platforms like greenhouse where roles may have already been setup.

Research

To make sure we’re building the best experience possible for our job seekers, and to improve upon what they are already using out there, we ran UXR sessions with 7 participants. The participant pool was made up of a mix of individual contributors from juniors to more senior IC’s and 2 people managers. Roles were primarily software engineering based since that’s our majority on the platform, with 3 candidates in design and product.

We learned pretty early on that participants disliked reading through job posts, and find it hard to find important information that helped them decide whether to apply or not. In terms of ranking, we learned that some of the most important pieces of information includes compensation ranges, workplace policy (Remote or Hybrid), required YOE, and finally tools that the role uses. This was especially important for engineers who need to know what coding language is used on the team. 


“I don’t tailor my resume for each job, instead just look for jobs that match my skills (java, docker, etc)”

Other learnings that were just as important were how happy people currently working at the company are, both on a company and team level.

“I wish there was a place where testimonials are there for teams. If there's a large number of happy people at a company, usually it's a good chance that the company is not toxic” 


Outcome

We launched our first version of the jobs feed that drove a 45% increase in job post views compared to our test feed that Blind Korea launched a few months prior to the US release date.

Have since added 20,000+ open roles on the platform.

Helped the company raise an additional 15m in funding due to the new jobs experience and business line.

Problem

Blind users are using the platform to compare job offers, market rate compensation, and research prospective companies. However at the time, Blind had a very simple and limited tool for salary comparison that did not encompass the complexities of comp packages at most companies. Furthermore, users would ask for compensation information on posts when it’ sometimes irrelevant due a lack of home for compensation and equity tools.

Opportunity

Due to the immense interest around compensation, there was a very direct way to connect the needs of ours users with a more robust compensation / equity comparison tool. This also provided a gateway for us to collect compensation information in a more systematic way; improving the quality and depth of the data we can show.

Research

Conducted 3 rounds of UXR with users of Blind for both the data visualizations as well as the compensation input flows. Generally participants found the data visualization very useful, and would use it to during job search for viewing comparable offers. Participants understood the give-to-get model from the start, and was ok providing their own information first. Similar to other services like Glassdoor and LinkedIn.

Some of the recommended changes and findings we iterated on included clearer message and controls for how users indicate vesting schedules during input flows. Allowing users to break down weighted vested schedules by percentages made a big difference comprehension. We’ve also simplified the page for filtering where all modules are filtered with controls only at the top. There are no filtering controls for specific modules. 


Outcome

Designed and shipped the initial total compensation tool within 4 months. Since then we’ve collected 15,000+ salaries from FAANG to startups over a period of 1.5 years. Our users have been not only using the data, but we’ve also publicly shared our data to the rest of the tech community via published infographic / posts for others outside of the platform to see.

The design language from this project was also applied to our core design system. Primarily the overhaul to the 8pt grid that was used along with additional components for filtering controls and data visualization templates.

Problem

Dasher and Consumer communication is key to successful deliveries. Among other thing, it primarily facilitates successful deliveries at difficult drop-offs, and sets expectations when order is getting delayed. At the time of the project, the entry point is in app however the actual communication happens outside the app via phone calls/text messages (actual phone numbers are masked by Twilio). This takes the experience outside our control and exposes our users to a variety of reliability issues that are inherent with those channels like message latency and no send / read receipts.

Opportunity

By creating an in app / native communication experience we can increase successful contacts rates between Dashers and Consumers which in turn can help reduce delivery errors like “never delivered”. By making it easier to initiate conversations between the two parties, there’s an opportunity to improve CSAT ratings, and reduce contact rates to our support teams as well. And lastly, by building the first ever chat system within DD, other teams across our marketplace can potentially leverage chat across their use cases as well ie: chat with support agents or merchants.

Research

Some of our key takeaways from usability testing included participants feeling that the new chat experience was positive and simple. All participants when shown pre-canned responses understood how they were used, and compared it to that of other services like messenger and gmail that uses a similar prompt.

Participants also mentioned that finding the entry point for chat was easy, and that it was contextually closer to where they were looking for their Dashers information on the tracker card.

“Usually I receive texts to my direct cell number from the drivers random number, I much prefer having the convo happen in the app.”

“I will be very comfortable sending an image that would make the delivery process easier. I know that my personal information is protected because I am communicating with the dasher via the "Doordash" app so it is secure and safe to use.”

Results

The feature drove a clear reduction in our primary metric - Never Delivered Rate decreased by 0.05% or 11.25% relative. This translates to $11.5M in annualized cost savings.


This was primarily driven by creating an easier communication experience for Dashers and Consumers and a faster, more reliable communication channel


The percent of text messages successfully delivered increased from 83.3% to 93.8%, and the percent of delivered messages with < 1 minute of latency increased from 92.4% to 99.0%

As a platform owner for delivery and quality, I’ve created guidelines and pattern documentations for the design team to follow. These documents help guide teams who need to build new features into the delivery side, understand how deliveries work, and what types of patterns are available to use, along with rules on why certain components / elements are where they are on the page.

Delivery Timeline Audit

Due to the delivery timeline being different across the different platforms we operate on (iOS, Android, Web), I put together an audit with the engineering team to pull out delivery status’s starting from the moment you place the order, to when the order is completed. The completed audit, was then used by designers and product managers who need to use the delivery status chart either as guidance for when to pop-up certain notifications, messaging, or even to add new status’s in a way that works with timeline as it stands.

Delivery Tracker Guide

As the the need to have more features added into the post delivery screen grew, I created the tracker card guide as a way for designers building onto the tracker card to better understand the in’s and out’s for how the card segmentation is broken down. Within the guide is general guidelines on where each segment of the tracker information sits, the rationale behind the ordering, and detailed views of the different states of each segment. Designers could then play around with the segments themselves, when exploring solutions, while knowing where the boundaries are. This prevents teams from breaking any fundamental patterns that could effect the usability of the tracker card.

Support Page Templates

Once all the updated support flows were updated with the latest resolution / content, I started putting together a template to help other designers looking into building on top of the support framework. This guide helps incoming product and design functions understand where order issue collection starts, how we ask for details for the defect, and finally what types of outcomes are possible based on the defects currently in place. The goal was to make it easier for other marketplace teams like merchant and dasher, to create support flows of their own, while still adhering to a set of consistent patterns.

Goal

To build on the success of video in its first year and 2x its growth in 2018. The goal is to expand the format’s ability to help advertisers tell their stories, with an explicit sub-goal of capturing brand awareness budgets.

Problem

We believe video on Pinterest must transform significantly to achieve our goals. There are three axes on which video could improve significantly:

Surfaces - Due to insufficient and limited placement, it is difficult to elevate premium video content over other videos for brand concerned advertisers.

Pin rep - Our basic square format forces widescreen videos to be cropped and discourages the creation of vertical videos, currently the most popular format on mobile.

Behaviors - Video capabilities on the platform are limited to a “play, pause” modality, making it not very actionable for Pinners.

Process

- Pre project kick off explorations

- Early concepts to help guide xfn team on feature framework

Goal

Build a scalable social camera on Messenger that enables people to express themselves via content sent to thread or to stories.

Problem

Phase 1:

The inherited problem with the messenger camera revolves around the clarity / placement of tools and a system that doesn’t scale well with new tools being added in.

Scalability - Messenger camera started out with limited and simple editing tools. However, as the amount of features and tools grew, we needed a better IA that could grow with the tools that we are introducing.

Phase 2:

Alignment with FB product suite - After the first design and interaction study, we shifted into making sure that the camera was aligned with how people expected to use other social cameras within the FB family of apps. Since Instagram’s camera was what people were used to most for content posting and story creation; we aligned parts of the camera to bring back a sense of familiarity between the suite of apps.

Camera capture clean up - We realized that the second iteration of camera wasn’t giving the new camera modes the highlight and room it needed to perform. This was due to the controls and mask effects competing for attention, so we began by cleaning up the view on the initial capture screen through hiding the masks under a button / tap.

Results

Phase 1 testing -

image editing from thread view went up 3.2%

Post capture AR mask usage up 2%

Phase 2 solution -

Finished camera redesign alignment with other FB social cameras ie: IG

End to end camera flow that incorporates iOS system components when possible to reduce the use of custom components and app size. The new camera design is now integrated into the larger Messenger Lightspeed rewrite to improve performance and reduce app size overall.

Process

- IA audit

- Interaction study

- cross product alignment

- usability study

One tap is the bread and butter ad format for direct response advertisers on Pinterest. These ads are made to efficiently drive Pinners to the Partners site as quickly as possible, while still providing value up front to inspire Pinners to take action.

Problem 

The first version of one tap had a few issues with the experience for both Partners and Pinners. On the Pinner side, the creative was lost and taken away when the site loaded into view. This made it difficult for Pinners to really look and evaluate the creative that they are used to viewing on a regular closeup. For Partners, they were seeing a dramatic increase in CTR, however, conversions were neutral to down. The theory is that Pinners were clicking in, and backing out instantly due to the poor transition from home feed to the site. 

Goals

- Increase click-throughs

- Provide a better viewing experience for Pinners when transitioning into the site 

- Define a better success metric that accurately logs high intent clicks into the site

Process

V2 - In this version, the ad creative stays on click, resting right above the site. Pinners can now view the image for as long as they want, and when they're ready, scrolling up would take them directly to the site. 

Metrics changed from direct click-throughs to gCTR (good click-throughs). Good clicks were defined as 3 seconds or more on the site, which helped push the quality of the format, along with moving away from empty clicks. 

V3 - The third version of one tap was focused on bringing a more immersive and premium feeling to the format, along with driving up good clicks. This was done by keeping the website in view, no matter the height of the ad creative. Depending on the height of the image, the website would either be shown in a larger view or stay sticky as a footer at the bottom. 

Results

Click-throughs went down, however, gCTR went up by 2% along with conversions. This essentially meant that we were getting more qualified clicks. There were less Pinners going into the site, however those who went into the site spent longer sessions browsing and converting more often. 

The project was to turn Pulse from a news aggregator to a more relevant and personalized stream of content. Instead of displaying a long condensed list of articles to sort and parse through, the idea is to focus on larger more visual content pieces that show reasons for why someone is seeing what they are seeing. The reader page has also been redesigned with serifed type faces, and larger cover photo experiences to help bring each article to life. Some of the bonus features in the redesign, includes an easy to tap reader (scraped) view that takes 3rd party articles and scrapes the text into our own template design. 

Process: 

The lead designer and I split up the work between the product, where I focused primarily on the reader page design. Along with owning the reader page, I assisted with feed / header transitions and design alignment. One of our major goals during the redesign was to also align our animation and design elements with the rest of Linkedin's content products.  

Deliverables: 

-Reader page interaction patterns

-high fidelity reader page mocks and specs 

-interaction prototypes for feed and reader page

-design sync of mobile reader page with desktop

Helping users complete their profiles to get better engagement with recruiters on the platform.