Things I've built

Portfolio

Products I've led at work with big teams, next to side projects I've built on my own, where Claude wrote the code and I made the calls.

Products I've led

Ontra's Markup Builder: AI-suggested edits to a non-disclosure agreement, shown next to the firm's playbook — Image: Ontra

Ontra Contract Automation

VP of Product / GM · 2026

Tools that take routine contracts, like NDAs, from the moment they arrive to signed, for private-markets firms and the attorneys who negotiate them.

The problem: Private-markets firms get a steady stream of NDAs and other routine agreements. Each one has to be read, marked up, negotiated with the other side and signed, and the back-and-forth slows down the deal it's attached to.
What I focused on: Streamlining the whole negotiation, start to finish, with tools for two groups: the attorneys reviewing and negotiating each contract, and the people at the firms who send contracts in and need to know where they stand. Better AI markups were one piece. I set up a weekly loop where our legal experts scored the output alongside the engineers, so we measured quality instead of debating it. I also stood up a six-week team to rebuild the product around agents, which pulled roadmap work forward by years.
My role: I owned the product line as GM, a 9-figure business, and led its product, design and engineering teams.

Tilt's HR dashboard: leaves waiting on HR or the employee, total leaves by type and leaves by stage — Image: Tilt, HR dashboard, 2025

Tilt leave platform

VP of Product · 2023 – 2026

Leave management for HR teams and for the employees going through a leave.

The problem: A leave, whether it's a new baby or a health scare, is one of the most stressful stretches of someone's working life. The rules change by state and even by city, so HR teams get pulled between getting compliance right and actually being there for the person.
What I focused on: Making the routine parts automatic so HR could spend their time on the human parts. I led the company's move to AI, from the infrastructure to how we built software to our first AI features for customers. I also brought in product operations, so decisions got made the same way every time.
Results: Gross margin rose 30+ points to SaaS-standard levels while ARR kept growing well ahead of SaaS benchmarks.

Degreed Plan Insights reporting from 2022: skill-rating charts against target levels, and how many followers started and completed each pathway — Image: Degreed help center, Plan Insights, 2022

Degreed admin & reporting

Product Manager, then Director of Product · 2019 – 2023

The admin side of Degreed's learning platform: settings, reporting and the tools companies use to run learning for their people, customers and partners.

The problem: Large companies run Degreed for huge workforces. Their admins need to set it up, prove it's working and sometimes run it for outside audiences too, and every gap in the tools turned into a ticket for our team.
What I focused on: Self-serve. I came from client success, so I knew which requests ate our team's week. We turned those into admin tools and in-app reporting, which cut the manual work and improved gross margin.
My role: I started as the PM for admin features and internal tools. As Director, I coached a team of PMs across reporting and insights, internal tools, extended enterprise and admin settings.

Side projects

Next Chapter

Side project · 2026

Checks job postings against what you care about and shows you the evidence. Invite-only beta.

Why I built it: When I was looking at senior product roles, most of my time went to reading postings just to see if they were even close. The things I cared about most, like pay, travel and whether I'd manage a team, were often buried or not mentioned at all. I tried Teal, Huntr and Jobscan. They're good at helping you apply to more jobs. I wanted help applying to fewer.
The bet: I'm betting that senior people will trust a tool that shows its work. Next Chapter checks a posting against each of your priorities one at a time and quotes the line it's going off. If the posting doesn't say, it tells you so and suggests what to ask the recruiter. That's a smaller audience than everyone looking for a job, but I think it's the group that cares most about getting this right.
How I'm testing it: I used it every day in my own search for a few weeks before showing anyone. Then I opened a small invite-only beta. I'm paying for the AI for now, with daily and monthly caps, because first I want to know if people keep coming back. Pricing can wait. I did work out the costs early: about a penny to check a posting and about 50 cents to search the open web.
Still figuring out: Exactly who this is for, and what they'd pay for it. That's what the beta is for.

nextchapter.jeffokita.com →

Wishful

Side project · 2026

Wish lists for families. Add gifts, rank them and share one link.

Why I built it: A few times a year my mom asks what I want for a birthday or the holidays, and I send her a Google Sheet. It mostly works. But nobody can tell what someone else already bought, and "really want" and "nice to have" look the same in a spreadsheet. There are plenty of wish list apps out there, and most families I know still use a doc or a sheet anyway.
The bet: It only had to be easier than the spreadsheet, for the person making the list and for everyone buying from it. So guests never need an account. My parents shouldn't have to sign up for anything to buy a gift. I also timed it for the holidays, when people are actually making lists, and every list someone shares shows the app to a few more families. 500 Startups taught me to think about distribution from the start, and that matters even more now that anyone can build an app in an afternoon.
How I'm testing it: The first version took about four hours, so being wrong wouldn't cost much. I added a feedback button and a simple usage page right away to see what people actually did with it. I'll find out in December whether families stick with it or go back to the spreadsheet.
What I learned: My best calls were things I took out. I'd built a toggle that let you peek at what had been claimed, and I removed it because one tap could spoil the surprise. I skipped product search and payments too. The behind-the-scenes stuff, like storing data, sending email and setting up the domain, caused more trouble than any feature did.

wishful.jeffokita.com →

Reading Shelf

Side project · 2026

The books on my to-read pile. Friends and family vote on what I read next.

Why I built it: I have a stack of books I want to get to and can never decide what's next. Asking in a group text gets a reply or two.
The bet: If voting took a few seconds, more people would do it. The first version had you drag the books into order, which was too much work. I switched to a +1 button you can tap on as many books as you like. There's no sign-in, your name is optional and only I see the results, so nobody feels awkward telling me to skip one.
How I'm testing it: There's a waitlist at the end for anyone who wants a shelf of their own. If enough people sign up, I'll build that next.

Visit Reading Shelf →