Harshil Thakkar

• +10% — Transaction Volume
• $400,000+ — Monthly BNPL Transaction Volume
• -12% — Fraud Reduction
• -15% — Compliance Escalations
• +20% — BNPL Adoption

Digital banking users lacked real-time visibility into payment status, settlement timing, and FX handling across card and A2A payment rails. Treasury teams struggled to reconcile transactions quickly due to fragmented data across systems. Payment risk controls and fraud detection were also reactive, leading to increased compliance escalations and operational overhead.

SOLUTION
• Implemented a real-time payment visibility layer enabling customers to track transaction status, settlement updates, and FX details directly within digital banking channels.
• Introduced ML-based fraud detection integrated into payment authorization flows with configurable risk thresholds and step-up authentication.
• Built unified transaction monitoring and treasury reconciliation capabilities across card and A2A payment rails.
• Enabled scalable infrastructure supporting new financial products such as BNPL within the digital banking ecosystem.

I increased digital payment transactions by 10% through real-time transaction visibility and reduced fraud by 12% and compliance escalations by 15% using ML-based risk controls. I achieved $400,000+ monthly BNPL volume with 20% adoption growth.

• +25% — Ecosystem Partner Growth
• +18% — A2A Transaction Volume
• -20% — Integration Support Overhead
• -15% — Compliance Escalations
• -30% — Partner Onboarding Lead Time
• 100% — Regulatory Audit Compliance Rate

Open banking regulations required banks to expose payment and account capabilities to 3rd-party providers through regulated APIs. Existing internal APIs lacked versioning, consent boundaries, and behavior guarantees required for external use. We needed to enable fintech innovation without compromising security, customer trust, or regulatory compliance.

SOLUTION
• Architected a governed open banking API platform enabling 3rd-party providers to access payment and account capabilities through regulated interfaces.
• Standardized API contracts with clear versioning, consent models, and access scopes aligned with compliance expectations.
• Designed integration patterns that supported external developer adoption while maintaining auditability, security, and operational control.

I expanded the fintech partner ecosystem by 25%, enabling scalable 3rd-party integrations. I also increased A2A payment volume by 18% via open banking APIs and reduced partner integration time by 20% and operational errors by 15%.

• -35% — Manual Analysis Effort
• +18% — Revenue Uplift
• +6% — Authorization Approval Rate
• -10% — Fraud Loss Rate

Authorization strategies were tuned manually using historical reports and lagging indicators, and static rules struggled to adapt to evolving payment behavior, merchant models, and cross-border transaction patterns. Small configuration changes could materially affect approvals, revenue, and fraud exposure across high-volume payment systems.

SOLUTION
• Built an AI-assisted authorization strategy optimization layer, analyzing historical payment outcomes to identify improvement opportunities.
• Modeled approval, fraud, and customer experience trade-offs to simulate alternative authorization strategies.
• Generated contextual recommendations for rule thresholds, strategy ordering, and merchant-specific configurations.
• Designed governance controls with confidence scoring, impact estimates, and human approval before activation.

I increased authorization approval rates by 6% while maintaining fraud controls. I also reduced fraud loss exposure by 10% through targeted strategy optimization and manual analysis effort by 35% and improved payment revenue by 18%.

• +40% — Early Risk Detection
• -22% — False Positive Alerts
• -30% — Investigation Time
• +18% — Operational Efficiency

Fraud systems focused on known bad patterns and real-time enforcement, leaving emerging behavioral shifts unnoticed. Risk teams lacked tools to interpret unusual transaction patterns across customers, merchants, and corridors, and anomalies were often buried in logs or triggered unnecessary enforcement actions, increasing customer friction.

SOLUTION
• Built an AI-based anomaly detection platform analyzing transaction, customer, merchant, and corridor behavior to identify contextual deviations.
• Established adaptive baselines using historical patterns to detect meaningful anomalies rather than global outliers.
• Designed explainability features highlighting deviation drivers such as temporal shifts, peer comparison, and behavioral changes.
• Positioned the platform as decision intelligence supporting risk teams rather than automated enforcement.

I improved early detection of behavioral risk signals by 40% across payment activity, reduced false positive alerts by 22% and investigation time by 30%, and increased operational efficiency by 18% through explainable anomaly insights.

• +12% — Listing Conversion Rate
• -18% — Time-to-price Decision
• +21% — Pricing Accuracy
• +27% — Insight Adoption Rate

Real estate users faced abundant predictive signals but lacked clarity on how to interpret them in decision contexts. Conflicting indicators across pricing models, buyer demand signals, and market trends created uncertainty. Users also often ignored analytics or misinterpreted signals, reducing the value of the platform’s predictive capabilities.

SOLUTION
• Built an AI-driven insights and recommendation layer, translating predictive models and market signals into contextual decision guidance.
• Prioritized signals based on user role, property state, and market dynamics to surface timely insights.
• Generated short explanations connecting recommendations to observable signals, improving trust and interpretability.
• Embedded insights directly within property views and pricing workflows to ensure guidance appeared at decision moments.

I improved listing conversion by 12% through AI-driven pricing insights, reduced pricing decision time by 18% via contextual recommendations, and increased pricing accuracy by 21% and insight adoption by 27%.

• +21% — Pricing Accuracy
• +17% — Buyer Conversion
• -19% — Time-to-price Decision
• +26% — Market Insight Adoption

Real estate pricing relied heavily on lagging comparable sales and human intuition, and static valuation approaches struggled in thin markets and rapidly shifting demand conditions. Users needed forward-looking price guidance that reflected market dynamics while remaining interpretable and trustworthy.

SOLUTION
• Built AI-based property price prediction models combining comparable sales data with behavioral and market signals.
• Incorporated factors such as listing velocity, demand intensity, seasonality, and price change behavior.
• Designed predictions as probabilistic price ranges with confidence bands rather than single estimates.
• Paired predictions with contextual explanations to support buyer, seller, and agent decision-making.

I increased pricing accuracy by 21%, improving buyer and seller decision confidence. I also improved buyer conversion by 17% through clearer market pricing signals and reduced pricing decision time by 19% via predictive price guidance.

• +29% — Property Engagement
• +22% — Market Insight Adoption
• -17% — Decision Time
• +19% — Listing Conversion

Buyer demand signals existed only as raw behavioral events such as views, saves, and inquiries. High traffic did not always reflect actual demand, making it difficult for agents and sellers to accurately gauge interest. Additionally, traditional indicators relied on closed transactions, which lagged real market behavior by weeks or months.

SOLUTION
• Built an AI-based buyer behavior analytics platform, translating engagement patterns into demand intelligence signals.
• Modeled buyer intent using engagement depth, recency, repeat activity, and cross-property comparisons.
• Created contextual demand indices comparing behavior against historical baselines and similar listings.
• Designed insights to emphasize trend direction and market momentum rather than isolated activity spikes.

I increased property engagement by 29% and market insight adoption by 22%. I also reduced decision analysis time by 17% through behavioral demand signals and improved listing conversion by 19% as users better understood buyer interest.

• -12% — Fraud Loss Rate
• -15% — Compliance Escalations
• +9% — Authorization Approval Rate
• -20% — False Positive Rate

Payment risk systems struggled to adapt to new transaction patterns across BNPL, tokenized payments, and digital wallets. Static rules created excessive false positives, while evolving fraud tactics increased risk exposure. Risk decisions also needed to remain explainable and predictable while operating at real-time authorization scale.

SOLUTION
• Integrated AI-assisted fraud signals into real-time payment authorization flows.
• Combined probabilistic risk models with deterministic rules to balance adaptability and stability.
• Designed explainable risk outputs so operations and compliance teams could understand flagged transactions.
• Implemented controlled threshold tuning and evaluation windows to maintain predictable system behavior.

I reduced fraud loss rate by 12% while maintaining stable payment authorization behavior. I also reduced false-positive transaction declines by 20%, increased authorization approval rates by 9%, and lowered compliance escalations by 15%.

• +11% — Transaction Success Rate
• -28% — Payment Failures
• -32% — Incident Resolution Time
• +19% — Platform Reliability

Payment routing logic became increasingly complex as transaction volume, integrations, and payment methods expanded. Failover paths and retry behavior were inconsistent, causing unpredictable outcomes during outages. Routing decisions also needed to maximize transaction success while preventing cascading failures or latency spikes.

SOLUTION
• Designed a resilient payment routing framework directing transactions across processors and networks.
• Standardized routing intent, fallback paths, and controlled retry logic to prevent cascading failures.
• Implemented explicit failover behavior distinguishing transient errors from systemic outages.
• Improved routing observability so operations teams could understand routing decisions during incidents.

I increased the transaction success rate by 11% during peak traffic conditions and cut payment failures by 28% through controlled routing and failover logic. I also reduced incident resolution time by 32% while improving platform reliability by 19%.

• +24% — Digital Wallet Adoption
• +18% — Mobile Payment Transactions
• -21% — Checkout Friction
• +14% — Payment Approval Rate

Mobile wallet adoption accelerated, requiring the platform to support tokenized transactions and wallet provisioning flows. Integrations also had strict certification requirements from wallet providers and card networks, and wallet functionality needed to integrate into existing payment infrastructure without fragmenting core payment logic.

SOLUTION
• Introduced digital wallet support by extending the core payments platform rather than creating parallel transaction flows.
• Integrated tokenized wallet transactions into existing authorization and risk decisioning paths.
• Sequenced certification-driven platform changes across provisioning, encryption, and authorization layers.
• Coordinated internal teams and external wallet providers to meet launch requirements without destabilizing the platform.

I increased digital wallet adoption by 24% across mobile payment channels. I reduced checkout friction by 21% via streamlined wallet-based payments and improved mobile payment transaction volume by 18% while maintaining stable authorization behavior.

• +18% — Checkout Conversion Rate
• +22% — Average Order Value (AOV)
• +24% — Merchant BNPL Adoption
• +32% — BNPL Transaction Volume
• -17% — Payment Processing Latency

BNPL introduced credit decisioning and installment logic directly into high-volume checkout flows, creating new approval paths and failure modes. Rapid adoption risked destabilizing existing payment infrastructure and degrading checkout reliability. Eligibility decisions also had to remain fast, deterministic, and compliant as transaction volumes and merchant demand increased.

SOLUTION
• Integrated BNPL flows into existing authorization and routing infrastructure rather than building a separate payment system.
• Designed deterministic eligibility and decline handling to avoid broken checkout states and reduce merchant support issues.
• Sequenced rollout and scaling carefully to maintain platform stability while supporting rapid adoption and higher transaction volumes.
• Introduced guardrails to balance conversion growth with credit risk and system performance.

I scaled BNPL to $400,000+ monthly volume without core system downtime. I boosted conversion by 18% through frictionless credit approvals and maintained 99.9% reliability during rapid merchant adoption.

• +28% — Incentive Program Volume
• +99.7% — Payout Accuracy
• +15% — Margin Protection
• -25% — Manual Reconciliation Effort
• +41% — Rebate Program Adoption

Rebate programs involved tiered thresholds, cumulative purchase volumes, and retroactive adjustments across long contract periods. Manual spreadsheets also created reconciliation issues, inaccurate accrual estimates, and frequent partner disputes. The platform needed to align commercial incentive structures with financially accurate, auditable calculations trusted by finance and partners.

SOLUTION
• Designed the platform as the system of record for rebate programs rather than a reporting layer.
• Introduced declarative rebate rule definitions to ensure consistent, auditable calculations across programs.
• Separated earned rebates from paid rebates to enable accurate accrual tracking over time.
• Prioritized common rebate structures while controlling configurability to prevent unpredictable outcomes.
• Centralized progress tracking, accrual reporting, and payout calculations into a single platform.

I automated 16+ million annual claims with 99.7% accuracy, reduced manual reconciliation by 25% via automated accruals, and slashed partner disputes by 30% through real-time visibility.