cover_letter / art_x5w0amqX2cU

Dear Plaid Hiring Team, Plaid sits at the infrastructure layer of the modern financial system — the connective tissue that lets millions of people move money, access credit, and manage their financial lives through the apps they already use. That mission resonates with me directly: at Fintellect AI, I built a platform specifically to close the financial literacy gap for retail investors, and the friction I encountered in accessing reliable, timely financial data made Plaid's network feel less like a vendor and more like a prerequisite for the category. I'm applying for the Product Manager, Data Supply role because the problem you're describing — allocating constrained provider capacity, managing traffic health at scale, and building the operating framework where none yet exists — is exactly the kind of foundational infrastructure challenge I've spent my career gravitating toward. **Technical and Analytical Foundation** My technical credibility starts early: I hand-coded backpropagation through time in C++ at UC Berkeley in 2004, published at NeurIPS in 2014 on neural networks for protein structure prediction, and in 2026 rebuilt that system in PyTorch spanning 413 parameters to 8 billion — a 19-million-fold scale increase. That arc matters here because reasoning about large-scale systems, load behavior, and trade-offs between throughput, latency, and cost is not new territory for me. At Intuit, I owned the ICE platform — the internal compute and event infrastructure underpinning QuickBooks, TurboTax, Mint, Mailchimp, and Credit Karma. I drove 275% YoY growth in ICE engagements, scaling to 675M+ in FY23, and led the throughput migration from 6K to 50K TPS via rSocket, supporting approximately 1.5M concurrent connections at sub-25ms TP99. That work required exactly the systems thinking Plaid is asking for: modeling capacity constraints, managing load across products with competing priorities, and translating engineering trade-offs into product decisions that held up under scrutiny from engineering leads and the CTO alike. I also worked directly with SQL and BigQuery telemetry to prioritize developer pain points across ~20 mobile apps and 30+ product SKUs — the kind of data-grounded decision-making this role demands. At Splunk, I owned Search Service (Go microservices), Search Catalog (PostgreSQL metadata), and the Splunk Processing Language — building roadmaps and acceptance criteria for Splunk Cloud Services while leading a query performance initiative that achieved up to 10x improvements for a beta enterprise customer. I designed a repeatable RICE-based prioritization framework across three microservice backlogs, balancing internal partner, third-party developer, and Fortune 500 customer requirements simultaneously. **Why This Role, Why Now** The Data Supply Traffic and Health team is being asked to build an operating framework from scratch — defining how Plaid allocates constrained provider capacity, tracks cost efficiency across paid-access traffic, and balances data freshness against provider load. That is not a role for someone who needs a playbook. My career has repeatedly put me in exactly that position: building the ICE Self-Service platform from the ground up (reducing developer onboarding from weeks to minutes), launching Fintellect and Vantage as 0-to-1 products with no prior template, and establishing the MSaaS Drift Detection program at Intuit where I wrote the Java JAR library, designed the DevPortal UI, and built the remediation roadmap before any of those pieces existed. What specifically excites me about this role is the intersection of three things I find genuinely interesting: traffic allocation as a resource economics problem, the bridge function between data-provider relationship teams and internal product teams, and the opportunity to establish the metrics Plaid uses to run and improve this area long-term. The JD's mention of evolving ML models as a lever for traffic optimization also connects directly to my RL Workbench work, where I benchmarked GRPO, DPO, PPO, and nine other algorithms across TRL, VeRL, OpenRLHF, and NeMo RL — building the intuition for how model behavior changes under different training and inference constraints. **Selected Relevant Experience** - **ICE Platform Scale (Intuit):** Scaled ICE engagements to 675M+ in FY23 with 275% YoY growth; led rSocket migration from 6K to 50K TPS supporting ~1.5M concurrent connections at sub-25ms TP99 — directly analogous to managing traffic volume and provider load at Plaid's scale. - **ICE Self-Service Platform (Intuit):** Delivered DevPortal, GitOps config, and ICE Playground, reducing developer onboarding from 2–3 weeks to minutes in pre-prod and under 24 hours for production, while mitigating $1M+ in projected opex growth — demonstrating cost-efficiency framing alongside platform delivery. - **Enterprise Service Language Assessment (Intuit):** Conducted analysis across 9 languages using usage data and developer feedback to inform strategic investment decisions presented to the CTO — the kind of data-to-executive communication this cross-functional role requires. - **MSaaS Drift Detection Program (Intuit):** Initiated and built a Java JAR library to scan Git repos for configuration drift, partnered with Design on DevPortal UI, and built the remediation roadmap using OpenRewrite — a 0-to-1 initiative with no prior framework. - **Search Performance Optimization (Splunk):** Led query performance initiative for a beta enterprise customer, building a mirrored topology for benchmark testing and achieving up to 10x performance improvements — analytical, systems-level problem-solving under real customer constraints. - **RICE Prioritization Framework (Splunk):** Designed repeatable framework across three microservice backlogs, balancing internal, third-party developer, and Fortune 500 requirements — the kind of structured prioritization needed to allocate constrained capacity fairly across Plaid's products. - **Telemetry-Driven Prioritization (Intuit):** Worked directly with SQL and BigQuery usage data across ~20 mobile apps and 30+ product SKUs to surface and prioritize developer pain points — the analytical rigor required to establish and track traffic and cost metrics from scratch. - **RL Workbench (2026):** Built a benchmarking platform across TRL, VeRL, OpenRLHF, and NeMo RL with standardized throughput, memory, and convergence metrics — relevant to the JD's explicit mention of ML model evolution as a traffic optimization lever. **Closing** Plaid's mission — unlocking financial freedom for everyone — is not abstract to me. I built Fintellect specifically because access to quality financial guidance has historically been gated by wealth, and I've seen firsthand how much the reliability and freshness of underlying data determines whether that guidance is trustworthy. The Data Supply Health and Traffic team is building the foundation that every Plaid product depends on. I'd welcome the opportunity to bring my platform infrastructure experience, analytical background, and 0-to-1 product instincts to that work. Thank you for your consideration. Sincerely, **O. Felix Amoruwa** famoruwa@berkeley.edu | 909-731-9011 | felixamoruwa.info