cover_letter / art_PtjrsJPgzVs

Dear Snowflake Enterprise Apps & Agentic Systems Hiring Team, Snowflake is doing something rare: treating its own enterprise as the proving ground for the agentic future it wants to sell to the world. Being customer zero for Snowflake technology — redesigning Finance, People, and Legal workflows from first principles rather than layering AI onto legacy processes — is exactly the kind of zero-to-one product challenge I have spent the last several years building toward. When I delivered the ICE Self-Service platform at Intuit, I did not optimize an existing onboarding flow; I eliminated it, compressing developer onboarding from two to three weeks down to minutes in pre-production and under 24 hours for production. That instinct — start with the desired outcome, decide what should disappear — is the same instinct your job description calls for. **Technical and AI Foundation** My technical credibility spans from the foundational to the production-grade. In 2004 I hand-coded backpropagation through time in C++ for a protein structure prediction system that became a NeurIPS 2014 accepted paper. In 2026 I rewrote that same system in PyTorch across five neural architectures (feedforward, GRU, Transformer, ESM-2, multi-task), scaling from 413 to 8 billion parameters — a 19-million-fold increase — with MLflow experiment tracking, Optuna hyperparameter optimization, and FastAPI serving across six Docker containers. On the agent and orchestration side, I built 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. I architected a RAG retrieval pipeline with ChromaDB, multi-provider LLM orchestration across Claude, GPT-4, and Gemini, fallback routing, structured-output validation, and token-budget optimization. I built aeval, a local-first model evaluation platform with adversarial safety testing, refusal detection, bootstrap confidence intervals, Welch's t-test, Cohen's d effect sizing, and CI/CD regression gates — exactly the kind of evaluation and guardrail infrastructure that production-grade agentic systems require. I also built an RL post-training workbench benchmarking GRPO, DPO, PPO, DAPO, and eight other algorithms across TRL, VeRL, OpenRLHF, and NeMo RL frameworks, with live SSE metric streaming and GPU passthrough in Docker containers. This is not background reading — it is hands-on exposure to the training and evaluation machinery that underlies the LLM and agent patterns your team will deploy. **Why This Role** The through-line of my career is building platforms that eliminate friction at the boundary between complex technical systems and the people who depend on them — whether those people are developers at Intuit, traders using Fintellect, or job seekers on Vantage. The Staff PM role on Enterprise Apps & Agentic Systems asks for that same boundary work, but applied inward: redesigning how Finance, People, and Legal teams get work done, using Snowflake's own data cloud as the substrate. The opportunity to prove agentic patterns in production for internal teams and then carry those learnings to customers is a product arc I find genuinely compelling. What excites me specifically is the mandate to own the product problem from discovery through value realization — not to manage an application or implement a vendor solution, but to decide what work should be eliminated, what should be redesigned, and what should be delegated entirely to software. The emphasis on responsible operation — SOX compliance, auditability, data privacy, and responsible AI guardrails for sensitive Finance and People data — maps directly to the compliance and governance work I navigated shipping App Store and Mac App Store products under Apple's strictest sandbox and IAP requirements, and to the data-privacy consent gating I built into both Fintellect and Vantage. **Selected Relevant Experience** - **ICE Self-Service platform (Intuit):** Delivered DevPortal, GitOps config, and ICE Playground — reducing developer onboarding from 2–3 weeks to minutes in pre-production and under 24 hours for production, while mitigating $1M+ in projected opex growth. Scaled ICE engagements 275% YoY to 675M+ in FY23 across QuickBooks, TurboTax, Mint, Mailchimp, and Credit Karma. - **OpenClaw multi-agent orchestration (StreamIO AI):** Implemented gateway protocol, subagent delegation, profile management, and session switching coordinating AI agent workflows across four enterprise domains — a direct analogue to the orchestration, tool integration, and exception-handling patterns required for production agentic enterprise systems. - **RAG + LLM orchestration (Fintellect AI):** Architected ChromaDB vector store retrieval pipeline with multi-provider LLM routing, structured-output validation, and token-budget optimization; shipped 13 specialized AI advisors delivering context-aware, domain-specific guidance. - **aeval evaluation platform:** Built CI/CD-integrated model evaluation with adversarial safety testing, refusal detection, statistical rigor (bootstrap CIs, effect sizing), and automated safety gates — the evaluation and guardrail infrastructure production agentic workflows require. - **Vantage AI Job Search platform (StreamIO AI):** Led 0-to-1 product strategy, AI development, and go-to-market execution for an iOS/macOS platform with 40+ agent tools, streaming agent chat over FastAPI, enforced mutation approvals, and async job polling — shipping from discovery through App Store launch and iterative adoption. - **Enterprise-wide Service Language Assessment (Intuit):** Conducted analysis across 9 languages, 20 mobile apps, and 30+ product SKUs using SQL and BigQuery telemetry to surface developer pain points and inform strategic investment decisions presented to the CTO — the kind of data-grounded, cross-functional influence work this role requires. - **Scheduler Service delivery (Splunk):** Owned and delivered end-to-end in approximately four months, enabling scheduled search capabilities for first-party applications and demoed at Splunk .conf19 — evidence of moving quickly from zero to production in a complex platform environment. **Closing** Snowflake's thesis — that the agentic enterprise is not a future state but something being built right now, inside Snowflake itself — is the right thesis, and the Enterprise Apps & Agentic Systems team is the right place to prove it. I bring 12 years of platform product ownership, hands-on AI and agent engineering, and a track record of shipping products that eliminate friction rather than automate it. I would welcome the opportunity to discuss how that background maps to the problems your team is solving. Thank you for your consideration. O. Felix Amoruwa famoruwa@berkeley.edu · 909-731-9011 · felixamoruwa.info