cover_letter / art_WB9u3KBkl-M

Dear Faire Hiring Team, Faire is doing something structurally important: using data and machine learning to give independent retailers the same discovery and supply-chain leverage that large chains have always taken for granted. That mission — leveling the playing field for local commerce through platform infrastructure — resonates directly with work I have spent the last several years doing: building the data pipelines, ML tooling, and developer platforms that make every downstream product team faster and more confident. When I read the description of this role, I recognized the exact problems I have owned before, at Intuit and in my own AI platform work. **Technical and ML Foundation** My ML platform work spans both research and production. In 2014 I published at NeurIPS on neural networks for protein structure prediction — work that began with a hand-coded BPTT implementation in C++ at UC Berkeley in 2004. In 2026 I rewrote that system as a full production ML platform in PyTorch, adding five neural architectures (feedforward, GRU, Transformer, ESM-2, multi-task), MLflow experiment tracking, Optuna hyperparameter optimization, FastAPI model serving, and Docker orchestration across six containers — scaling from 413 to 8B parameters. More directly relevant to Faire's needs, I built an RL post-training workbench that covers the full RLHF/DPO pipeline: a Reward Lab for designing and A/B testing reward functions across four datasets (GSM8K, MATH, HumanEval, UltraFeedback), a Playground for real TRL-powered GRPO/DPO training with live SSE metric streaming on Apple Silicon and CUDA, and an Arena for head-to-head framework benchmarking across TRL, VeRL, OpenRLHF, and NeMo RL with GPU passthrough in Docker containers. I implemented 12 RL algorithms with algorithm-specific metric profiles and standardized throughput, memory, and convergence benchmarking — the kind of structured evaluation infrastructure that maps directly to Faire's need to measure ML platform success through user outcomes rather than process metrics. On the data platform side, I built aeval, a local-first model evaluation platform with five core eval types, adversarial safety testing with refusal detection, bootstrap confidence intervals, Welch's t-test, Cohen's d effect size, and CI/CD integration with regression detection and automated safety gates — running on a FastAPI orchestrator, TimescaleDB, Redis job queue, and Next.js dashboard. This is the same discipline Faire needs: SLAs, pipeline health, and measurable quality gates on data the business depends on. **Bridge** My arc runs from ML research to developer platform infrastructure to 0-to-1 AI product leadership — and this role sits precisely at that intersection: treating data and ML platforms as products with real users, measurable adoption, and an operating model that scales beyond its initial domain. **Why This Role at Faire** What draws me specifically to this role is the combination of marketplace data complexity and platform operating model work. Faire's ML systems — search, recommendations, personalization — depend on upstream data quality in ways that are hard to reason about without owning both layers simultaneously. The JD's framing of "data trustworthy by default" and "ML capabilities easy to adopt and operate in production" is exactly the right product framing, and it is the framing I applied at Intuit when I built the ICE Self-Service platform and the MSaaS Drift Detection program. I am also energized by the explicit mandate to establish a durable product-platform operating model — roadmap, intake, adoption tracking, cost and efficiency as health metrics — that Faire can extend as the platform function grows. That is the kind of structural work I find most valuable to do. **Selected Prior Experience** - **ICE Platform — Intuit:** Delivered the ICE Self-Service platform (DevPortal, GitOps config, 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. Achieved 275% YoY growth in ICE engagements, scaling to 675M+ in FY23 across QuickBooks, TurboTax, Mint, Mailchimp, and Credit Karma; scaled throughput from 6K to 50K TPS via rSocket migration supporting approximately 1.5M concurrent connections with sub-25ms TP99. - **MSaaS Drift Detection — Intuit:** Initiated and owned the Drift Detection and Resolution program: wrote a Java JAR library to scan Git repos for configuration drift, partnered with Design on DevPortal UI, and built a remediation roadmap using OpenRewrite — a direct analog to Faire's data contracts and dataset ownership work. - **Enterprise Language Assessment — Intuit:** Conducted an enterprise-wide Service Language Assessment across nine languages, analyzing usage data and developer feedback to inform strategic investment decisions presented to the CTO — the same kind of cross-functional, data-driven prioritization this role requires. - **Search Orchestration — Splunk:** Owned Search Service (Go microservices), Search Catalog (PostgreSQL metadata service), and SPL/SPL2; delivered Scheduler Service end-to-end in approximately four months; led query performance optimization achieving up to 10x improvements for a Fortune 500 beta customer. Designed a repeatable RICE-based prioritization framework for three microservice backlogs. - **RAG and Multi-Agent Orchestration — Fintellect/StreamIO:** Architected a RAG retrieval pipeline (ChromaDB vector store) with multi-provider LLM orchestration (Claude, GPT-4, Gemini), fallback routing, structured-output validation, and token-budget optimization. Implemented OpenClaw multi-agent orchestration framework with gateway protocol, subagent delegation, profile management, and session switching — directly relevant to Faire's AI data agents vision. - **aeval Evaluation Platform:** Built local-first model evaluation with statistical rigor (bootstrap confidence intervals, Welch's t-test, Cohen's d), adversarial safety testing, data contamination detection via SHA-256 hashing, and CI/CD regression detection — the evaluation and observability discipline Faire needs across its ML platform. - **Logging-as-a-Service — Kaiser Permanente:** Led development and enterprise rollout of Splunk Logging-as-a-Service at 1.7 TB daily volume across 200+ internal enterprise customers, including demand forecasting and capacity planning across multiple datacenters — operational discipline at the scale Faire's data infrastructure requires. **Closing** Faire's bet is that better data and ML infrastructure, owned as a product with real users and measurable outcomes, is what allows local commerce to compete. I have spent my career building exactly that kind of foundational infrastructure — at Intuit at scale, at Splunk in search and data services, and in my own platforms where I have had to make every architectural and prioritization decision myself. I would welcome the opportunity to bring that experience to Faire's platform team. Thank you for your consideration. O. Felix Amoruwa famoruwa@berkeley.edu | 909-731-9011 | felixamoruwa.info