Proof, not promises. Selected builds.

The challenge

A fintech data company needed a market-first product to help banks and payment providers compare rates and fees across global payment networks - but nothing like it existed. No data infrastructure, no collection processes, no QA.

What I did

Designed the product architecture and data model, defined the data collection strategy across dozens of sources, and established QA processes from scratch - working with the client's engineering team to bring it to production. Data error rate was held below 2% - best-in-class in the industry globally at the time.

The outcome

The product became the company's #1 revenue driver, reaching 30%+ of total revenue within 2 years. Parallel enrichment of the B2C data layer added ~17% profit uplift.

The challenge

A fast-growing platform was running manual onboarding and verification at scale - slow, expensive, and leaking money through fraud.

What I did

Redesigned the entire onboarding process into a scalable IAM/KYC platform: ML-based verification, document recognition, image control, and a human-in-the-loop model for edge cases. Built the end-to-end solution including an internal product for the verification (human-in-the-loop) team and a full model training pipeline.

The outcome

87% of the verification process automated - with no degradation in onboarding funnel quality. Operational costs reduced by 60%. Fraud-related losses down by ~$250K/month.

The challenge

Partner mass payout operations were semi-manual, error-prone, and costly - no automated validation, no risk controls, no financial visibility.

What I did

Rebuilt the mass payout and financial operations platform: redesigned workflows, launched automated validation, implemented risk controls and penalty processing.

The outcome

Improved platform security and audit-ready financial controls.

The challenge

A high-growth EV company needed an entire customer support function built before product launch - online and offline, across markets, including physical operations.

What I did

Built the function from zero: processes, tooling, team structure, office setup, and operational infrastructure including lost & found workflows. Designed for scale from day one.

The outcome

A scalable support model ready to grow with the product - without linear headcount scaling. Full operational infrastructure in place before user growth hit.

The challenge

A fast-scaling organisation was running on broken governance - legal approval cycles stretched to 8–9 months, slowing down every team that needed a decision.

What I did

Established cross-functional governance and decision-making frameworks across product, engineering, finance, legal, procurement, and DevOps. Aligned stakeholders on process, ownership, and escalation paths.

The outcome

Approval timelines cut from 8–9 months to 2–3 weeks. Significant improvement in execution velocity and cross-team trust.

The challenge

Product teams were dependent on data and engineering resources for routine analytical tasks - slowing down decisions and creating constant bottlenecks. Documentation was manual, inconsistent, and disconnected from what was actually built.

What I did

Led an AI transformation of core product processes across two workstreams. First, self-service analytics: introduced NL→SQL tooling, a data catalog, and a knowledge base covering internal services enabling PMs to get the data they need without pulling in data or engineering teams. Second, development documentation and validation: implemented AI-generated documentation from the codebase and a mechanism to cross-check product requirements against actual implementation. Trained the product teams on working with AI tools and embedded them into everyday workflows - not as a one-off initiative, but as a lasting change in how the team operates.

The outcome

Faster decision-making across product teams. Meaningful reduction in dependency on engineering and data resources for routine analytical and documentation work.