01 / PRIZEPICKS / PLATFORM OPERATIONS
From tools
to systems.
I owned product strategy and the roadmap across operational tooling, settlements and scoring, support AI, and risk.
OPERATIONAL TOOLING / SELECTED EXAMPLE
HQ
One back-office app shaped around the work teams needed to do.
Built around how operators worked.
Teams were working in database-oriented admin screens built for a much smaller product. Access controls and performance needed to mature, and recurring member incidents often relied on one-off engineering scripts.
I led HQ’s product strategy, roadmap, and development. Operator research and workflow mapping with Support, scoring, game operations, fraud, and other teams shaped what we built.
HQ spanned 10+ operational surfaces and 100+ features, including member visibility, permissions, and resolution tools.
HQ / CONTEXT · ACCESS · ACTIONS
Give operators what they need to resolve a member issue.
- Context: member information and relevant history in one place.
- Access: 18 roles and 150+ permissions, with controls for sensitive information.
- Actions: permissioned, auditable tools for refunds, adjustments, and recurring remediation.
HQ / REPORTED OUTCOMES
Faster responses and resolutions.
RESOLUTION DASHBOARD
Give operators a way to make it right.
Recurring refunds, adjustments, and other remediation moved into operator tools.
SCORING & SETTLEMENTS
Settlement speed
Improve payout speed and accuracy together.
Why this mattered
Signal. Across five major sports, faster settlement was associated with more repeat entries and daily lineup activity. It supported an eight-figure modeled opportunity—not a realized result or proof that speed alone drove activity.
Constraint. The largest delay from game end to payout was waiting for results to become eligible to settle; processing performance also mattered at peak load.
Product decision. I led threshold and stability checks during games, then post-game buffers for close outcomes more exposed to score corrections. High-confidence picks could settle sooner while guarding against resettlements. We also improved peak processing with Infrastructure.
REPORTED SETTLEMENT RESULTS
Faster payouts, more automation, fewer corrections.
Latency and resettlements are relative changes; auto-settlement is a percentage-point increase.
SUPPORT AI
Scale support without losing quality.
Knowing the policy doesn’t answer “Where is my withdrawal?”
I owned the support AI product roadmap with Customer Service. We started where conversation volume was high and CSAT showed the experience needed work: accounts and payments first, then promotions.
Scoring and settlement questions came later because they depended on live games, refunds, adjustments, and scoring policy. We needed the data and guardrails in place before the assistant could handle them well.
Customer Service refined procedures and brought feedback from conversations. We used that alongside release metrics to decide what to improve next.
TWO DIFFERENT MEASURES
Share of all inbound conversations automated.
CSAT increased during the same period; this does not establish that automation alone caused the change.
RISK → ACCOUNTS
Risk work led us into Accounts.
Investigators had to piece together signals from more than ten vendor systems. We centralized them in one investigator view and shipped initial high-confidence decisioning; broader real-time risk capability remained future work. As we built the fraud platform, it became clear that risk decisions could not be separated from login, verification, restrictions, and recovery. The same team later expanded into Accounts, where I led a broader product strategy across Manage, Protect, and Recover.
Explore Accounts ↗