Built for in-house traders, then for 25+ client brands.
60-second summary
- Who:
- In-house traders, then the sportsbooks and prediction markets that build our APIs into their products.
- Key decisions:
- Build for the internal experts first; one shared capability instead of per-customer endpoints; automate a decision but keep its exceptions; fix data without changing what customers integrate against.
- Results:
- One trader per game instead of two, doubling in-play games offered; live CFB pricing from 1 live customer at the September 2025 launch to 7 by September 2026; optional filters used in 80%+ of alternate-line requests.
- My role:
- Roadmap, requirements, release acceptance and customer follow-through; engineering and data science built the services.
I started with Swish’s in-house traders. Now I own NBA and college-football pricing API roadmaps used by 25+ client brands. Since January 2025, I’ve led 30+ releases with data science, engineering, trading and customer success teams, from product scope through launch, integration and adoption.
2× in-play games offered, with one trader covering each game instead of two
25+ client brands using NBA and college-football pricing APIs
1 → 7 live CFB pricing customers, September 2025 launch to September 2026
80%+ of requests to the NBA/CFB alternate-line endpoints use the optional filters
Build for the internal experts first
I started with Swish’s in-house traders, turning their requests into changes to our trading platform. That work let one trader cover each game instead of two, so Swish could offer clients twice as many in-play games.
That work included a redesigned CFB trading view, a training sandbox and the CFB in-play view released in August 2025.
Now I own the NBA and college-football pricing API roadmaps, where I look for the shared capability behind any one client’s request. My scope includes roadmap priorities, API requirements, release acceptance and customer follow-through.
Launch a new API and get seven customers live
A major client brought pre-match pricing in-house, making live college-football pricing a commercial priority. I prioritized CFB In-Play, which supplies prices while games are underway, and defined the first useful release: stable market identities, availability and suspension states, results and a workable testing path.
I opened the Test API more than three months ahead of launch and worked with data science, engineering, trading and customer success on a staged rollout. One customer offered live markets at the September 2025 launch; adoption grew to seven live customers by September 2026.
Customers needed different market coverage before they could go live. I kept the first release focused enough for the launch customer, then expanded shared capabilities to support the next integrations.
- Test API (May 2025)
- staged rollout
- 1 live customer at launch (Sep 2025)
- 7 live by Sep 2026
What integration partners need to adopt it
The non-actionable Test API opened in May 2025. In August, the production endpoint opened with suspended markets for testing, ahead of the September launch. That gave integration partners a testing path before they needed to offer live markets.
By 2026, the Test API matched production request and response structures 1:1, using live data during events and historical games in the offseason. Partners could validate against the same contract they would use in production.
One shared capability instead of per-customer endpoints
Customers repeatedly requested common statistical thresholds instead of every closely spaced alternate line. I turned those requests into one set of optional filters across the NBA/CFB alternate-line APIs, defining stat-specific thresholds with customer feedback instead of a separate endpoint per customer. Engineering implemented the filters, and customers could request milestones or retain the full output. Filtering selects existing lines without recalculating prices.
The filters returned about 90% less alternate-line data and were used in over 80% of requests to the relevant NBA/CFB alternate-line endpoints.
Expand the shared capability
In a separate NBA case, restrictions blocked a client’s integration. I prioritized lifting those restrictions in the shared pricing endpoint instead of building a custom one. The client went live.
In January 2026, I led expansion of shared pricing and settlement support to 26 existing CFB team/match market types, preserving request and response structures. Other clients use that support, so the capability serves more than the original request.
Automate a decision, keep the exceptions
Two customers wanted faster CFB player-result confirmation. I defined a 30-minute rule for automatic confirmation of eligible player results after game completion, and led delivery with engineering through the existing results interface in September 2026. Players with recorded statistics were eligible; those without statistics still needed participation data, and later corrections remained possible.
A processing job checked eligibility every five minutes; team/match results were outside the change. I explained the participation exceptions, later corrections and need for continued polling to customers. Slow-confirmation complaints went from a couple per week to none reported from rollout through September 30, 2026.
Fix the data without breaking customer integrations
When two clients lost NBA live player selections, I traced the issue to a market-grouping and game-state mismatch. I made the case for a data-side fix that kept market IDs and client integrations unchanged. The fix shipped within a week, and the client confirmed the issue looked resolved.
Engineering implemented the fix; I followed the changes through the client’s confirmation.
Let’s talk.
I’m looking for product roles building AI for knowledge workers, working closely with customers, engineers and research teams to turn complex workflows into products people trust.
