I lead products, and I build them.
Eight years leading consumer products at TriPlay, including a marketplace for music rights that I'm taking to market now. Independently: Atlas Trade, a live product for organizers and participants, and two AI products, Ask Atlas and Career Copilot.
Team and company outcomes from commercial product leadership — not solo builds:
- Eight years in product leadership
- eMusic Live: concept to 100+ events
- Platform capacity scaled from 2K to 8K concurrent viewers before a major festival
- Multi-currency checkout expanded from 4 to 120+ currencies
- Search/discovery rebuild increased album purchases by 15% after a roughly 30% catalog loss
These were company products delivered with engineering, design, marketing, finance, customer success, and external partner teams — not solo builds. Together, they show both sides of product leadership: creating a new platform from zero and improving an established commercial product over time.
eMusic Live turned digital concerts into complete commercial events rather than simple video streams. Fans could buy access, bundles, VIP experiences, and merchandise; interact with artists during shows; and move through timezone-aware event journeys. Artists, promoters, vendors, and internal teams needed tools to configure, operate, fulfill, and measure each event.
My scope: Led product from concept through launch and ongoing operation across ticketing, global payments, merchandise fulfillment, artist tooling, fan engagement, event-page publishing, analytics, and platform reliability.
- Took the platform from concept to 100+ live events, coordinating cross-functional and partner teams across product definition, delivery, launch, and post-event learning
- Changed the roadmap after load testing exposed a 2K concurrent-viewer ceiling, built the case for reliability work, and led the team to 8K concurrent viewers with zero load-related downtime during a three-day festival
- Expanded global checkout from 4 to 120+ currencies while keeping prices stable for buyers and reconciliation manageable for Finance
- Built beyond video delivery: artist-fan interaction, mobile-first viewing, ticket and merchandise bundles, vendor order routing, self-service event-page creation, and behavioral analytics
eMusic was an established buy-to-own digital music service spanning responsive web and native iOS and Android applications. Product ownership covered the complete customer and commercial system: acquisition, subscription lifecycle, pricing, checkout, catalog discovery, recommendations, mobile-platform constraints, fraud prevention, and executive business intelligence.
My scope: Owned and improved the product across engineering, mobile, marketing, finance, support, data, and executive stakeholders — balancing customer experience, platform policy, revenue, operating cost, and a changing music catalog.
- Turned a roughly 30% catalog loss into a discovery rebuild that replaced dead-end searches with relevant alternatives and increased album purchases by 15%
- Redesigned cancellation around customer reasons and targeted alternatives, doubling subscription save rates while producing structured insight into why customers left
- Led pricing, acquisition, and conversion experimentation across subscription tiers, album landing pages, and checkout rather than treating growth as a separate marketing function
- Improved the operating system around the product: rules-based fraud controls, chargeback workflows, cross-platform feature strategy, App Store payment-policy workarounds, and recurring subscriber/revenue/churn reporting
Live Product · GitHub Repository
The community tool is seventeen years old. It works, and first-timers quit halfway. I kept the matching algorithm and rebuilt the flows around it.
Finding a game and deciding what you'd give for it turned out to be different jobs. Newcomers, veterans, and the organizer get different controls in the same product. Spring 2026: 86% of people who listed an item ended up in a trade. The prior NYC trade, on the old tool, was 63%, with different people and different games. After the Summer trade, a short survey changed the fix I was about to build.
Live Product · GitHub Repository
Groups struggle to pick the right board game for a specific table — different players, time budgets, and experience levels. Atlas Realms turns a natural-language situation into a ranked shortlist with explainable reasons.
Defining decision: Models judge where language interpretation matters; code enforces every mechanical constraint. Validation rebuilt around casual players after early feedback overrepresented power users.
- 5–10s end-to-end latency, down from 31–35s (~78% reduction through caching and architecture changes)
- ~$0.0016 per query, weighted across the recorded query mix (August 2026)
- 100% consistency on a 10-prompt regression suite (regression discipline, not a universal accuracy claim)
Ask Atlas is the recommendation engine. Atlas Trade is the other product on the same site.
PM job search mixes slow profile construction with fast per-role tailoring — and most AI resume tools hallucinate experience or silently drop gaps. Career Copilot separates knowledge building from per-job work and keeps every claim tied to approved evidence.
Defining decision: Verbatim evidence selection by ID with human approval — no hallucinated experience, no silent omissions, gap flags for every unsupported requirement.
- 13-JD regression set with ~9–11/13 directional agreement on founder-labeled ground truth
- 11 completed model responses compared on failure severity rather than benchmark averages
- ~$0.019 cold / ~$0.013 warm per tailoring, down from ~$0.042 through model selection and caching
Closed pilot.
Decode the problem before choosing the solution. Research the existing process and validate pain before defining what to build.
Use models for judgment, code for invariants. AI handles language understanding; deterministic code owns rules that must never fail.
Evaluate failure severity, not averages. Build regression suites around the failures that break user trust.
Unit economics are architecture. If model cost doesn't work at intended usage, the product design isn't finished.
Validate with the right users against real data. Power users and casual users diverge; mock data hides integration failures.
Scope discipline saves projects. Defend the core mission; change course when evidence wins.
Director of Product at TriPlay, bringing a music-rights marketplace to market. Atlas Trade, Ask Atlas, and Career Copilot are in use beside that. Open to senior product roles, and to building with someone whose problem is still a mess.
Email: asher@asheratlas.com
LinkedIn: linkedin.com/in/asheratlas
Portfolio: https://www.asheratlas.com
Location: New York, NY