cover_letter / art_7Y_h3dtStfs

Dear Wil Pong and the WRITER Product Team, WRITER is building the infrastructure layer for enterprise AI — not just a point tool, but the platform through which companies like Mars, Marriott, and Vanguard orchestrate AI-powered work at scale. That framing resonates with me directly: my career has been spent at exactly this intersection of platform infrastructure, developer-facing products, and AI systems that need to work reliably in production. When I read WRITER's mission to expand human capacity through superintelligence, it maps cleanly onto the work I've been doing — from scaling Intuit's developer platform to 675M+ engagements to building multi-agent orchestration frameworks from scratch. **Technical and AI/ML Foundation** My AI/ML work is not recent decoration — it runs the full length of my career. In 2004 at UC Berkeley, I hand-coded a neural network in C++ with custom backpropagation through time (BPTT) for protein secondary structure prediction. That system was accepted at NeurIPS 2014. In 2026, I rewrote it as a full production ML platform in PyTorch spanning five neural architectures (feedforward, GRU, Transformer, ESM-2, multi-task), MLflow experiment tracking, Optuna hyperparameter optimization, and FastAPI serving — scaling from 413 to 8 billion parameters, a 19-million-fold increase. More directly relevant to WRITER's work: I built a 3-phase RL post-training workbench covering the full RLHF/DPO pipeline — a Reward Lab for designing and A/B testing reward functions across GSM8K, MATH, HumanEval, and 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 (PPO, GRPO, DAPO, REINFORCE, REINFORCE++, RLOO, DPO, SimPO, IPO, KTO, ORPO, SPPO) with standardized throughput, memory, and convergence benchmarking. I also 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 regression detection — the kind of rigorous evaluation infrastructure that enterprise AI deployments require. On the multi-agent side, I designed and implemented OpenClaw, a multi-agent orchestration framework with a gateway protocol, subagent delegation, profile management, and session switching — coordinating AI agent workflows across real estate, insurance, health, and financial-markets domains. This is the same architectural problem WRITER is solving at enterprise scale. **Why This Role** I've spent the last twelve years moving between two modes that this role demands in equal measure: setting platform strategy at the staff level and rolling up my sleeves to ship. At Intuit, I owned developer frameworks and platform infrastructure for a suite serving QuickBooks, TurboTax, Mint, Mailchimp, and Credit Karma. As a founder, I've been the sole PM, engineer, and go-to-market lead on multiple AI products shipped to the App Store. WRITER's Lead PM role asks for exactly that range — multi-quarter roadmap ownership one moment, unblocking execution the next — and I've been living that rhythm. What specifically excites me about this role is WRITER's position at the enterprise AI orchestration layer. The JD calls out fluency with LLMs, agents, platform architecture, and data — these are not aspirational skills for me, they are the substance of my last several years of work. The opportunity to shape product strategy in partnership with Wil Pong across multiple squads and surfaces, translating company-level goals into sequenced, ambitious bets, is the kind of scope I've been building toward. **Selected Prior Experience** - **Intuit — 675M+ platform engagements, 275% YoY growth:** Owned developer frameworks and platform infrastructure across ~20 mobile apps and 30+ product SKUs; scaled ICE throughput from 6K to 50K TPS via rSocket migration supporting ~1.5M concurrent connections with sub-25ms TP99. - **Intuit — Developer onboarding from weeks to minutes:** 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. - **Intuit — Enterprise-wide language strategy:** Conducted a Service Language Assessment across 9 languages (Java, Python, Kotlin, Go, TypeScript, Scala, PHP, C++, Groovy), analyzing usage data and developer feedback to inform strategic investment decisions presented to the CTO. - **RL Workbench — Post-training platform:** Built end-to-end RL post-training workbench benchmarking GRPO/DPO across TRL, VeRL, OpenRLHF, and NeMo RL; implemented 12 RL algorithms with cross-tab workflow lineage tracking and standardized benchmarking. - **aeval — Model evaluation platform:** Built evaluation platform with adversarial safety testing, refusal detection, statistical rigor (bootstrap CIs, Welch's t-test, Cohen's d), and CI/CD regression detection — directly applicable to WRITER's need for trustworthy, enterprise-grade AI. - **OpenClaw — Multi-agent orchestration:** Designed gateway protocol, subagent delegation, and session management for coordinating AI agent workflows across multiple industry verticals. - **Splunk — Search platform and query optimization:** Owned Search Service (Go microservices), Search Catalog (PostgreSQL), and SPL/SPL2; delivered Scheduler Service end-to-end in ~4 months and achieved up to 10x query performance improvements for a Fortune 500 beta customer. - **Vantage — 0-to-1 AI product with 40+ agent tools:** Shipped iOS and macOS job search and interview prep platform with streaming agent chat over a shared FastAPI backend with 40+ agent tools, enforced mutation approvals, and async job polling — full 0-to-1 ownership from discovery through App Store launch. **Closing** WRITER's vision — expanding human capacity through superintelligence, grounded in enterprise trust — is the right framing for where AI is actually going. The companies that win in enterprise AI will be the ones that make it reliable, governable, and deeply integrated into how work gets done. I've spent twelve years building toward exactly this problem: from developer platforms at Intuit to RL post-training infrastructure to multi-agent orchestration. I'd welcome the conversation about how I can contribute to WRITER's next chapter. Sincerely, **O. Felix Amoruwa** famoruwa@berkeley.edu · 909-731-9011 · felixamoruwa.info