cover_letter / art_jDaQzVlpPME

Dear Glean Hiring Team, Glean is doing something genuinely important: turning the fragmented knowledge scattered across an enterprise's SaaS stack into a coherent, actionable intelligence layer. The shift from reactive search to proactive, context-aware insights for leaders and teams is exactly the kind of hard, high-value problem that drew me into platform product work in the first place — and it mirrors a challenge I faced directly at Intuit, where I had to synthesize telemetry and usage data across 20+ mobile apps and 30+ product SKUs to surface developer pain points that no single team could see on their own. **Technical and AI Foundation** My technical credibility spans the full stack of what Glean's Enterprise Intelligence work will require. On the AI side, I built aeval, a local-first model evaluation platform with five core eval types (factuality, reasoning, instruction-following, safety, code generation), adversarial safety testing with refusal detection, and statistical rigor via bootstrap confidence intervals, Welch's t-test, and Cohen's d effect size — all orchestrated through a FastAPI backend, TimescaleDB for time-series eval history, and a Redis job queue. I also built a full RL post-training workbench covering 12 algorithms (PPO, GRPO, DAPO, DPO, SimPO, and more) with live SSE metric streaming, cross-framework benchmarking across TRL, VeRL, OpenRLHF, and NeMo RL, and GPU Docker passthrough — work I've written about publicly. These aren't survey projects; they are production systems built to answer real evaluation and training questions at depth. On the retrieval and orchestration side, I architected a RAG pipeline at Fintellect AI with ChromaDB vector store, multi-provider LLM orchestration across Claude, GPT-4, and Gemini with fallback routing, structured output validation, and token budget optimization. I also designed and shipped OpenClaw, a multi-agent orchestration framework with a gateway protocol, subagent delegation, profile management, and session switching — enabling coordinated AI agent workflows across distinct industry verticals. These systems required exactly the kind of thinking Glean's Enterprise Graph demands: how do you route context intelligently, maintain coherence across agents, and surface the right signal to the right user at the right time? My NeurIPS 2014 paper on neural networks for protein secondary structure prediction — and the subsequent rewrite of that system from a hand-coded C++ BPTT implementation to a PyTorch platform spanning 413 parameters to 8B — reflects a long-standing commitment to understanding AI systems from first principles, not just integrating APIs. **Why This Role** The Enterprise Intelligence PM role at Glean sits precisely at the intersection where I've done my most impactful work: translating ambiguous, data-rich platform capabilities into concrete product experiences that change how people make decisions. At Intuit, I didn't just manage a roadmap — I built Asterias, a declarative asset lifecycle management platform with a GraphQL API, initiated a drift detection program with a Java JAR library that scanned Git repos for configuration drift, and conducted an enterprise-wide service language assessment across nine languages to inform CTO-level investment decisions. That combination of technical depth and strategic framing is what I'd bring to defining what Enterprise Intelligence should look like at Glean. **Role-Specific Connection** What excites me most about this role is the mandate to move beyond personal productivity and build intelligence products that give leaders and departments visibility into what is changing and where to act — a category that doesn't yet have a dominant playbook. Glean's Enterprise Graph and Personal Knowledge Graph are genuinely differentiated assets, and the opportunity to turn that context layer into proactive dashboards, recommendations, and workflow triggers for cross-functional decision support is the kind of 0-to-1 product challenge I've pursued repeatedly, from the ICE Self-Service platform at Intuit to the Fintellect Agents embedded in a mobile investing app. I'm also drawn to the lean team structure — I work best when I have real ownership and can move between strategy and implementation without a layer of abstraction in between. **Selected Prior Experience** - Delivered 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 — a 0-to-1 platform bet that required both strategic framing and granular workflow design. - 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. - Worked closely with telemetry and usage data (SQL, BigQuery) to prioritize developer pain points across approximately 20 mobile apps and 30+ product SKUs; built Asterias, a declarative asset lifecycle management platform with GraphQL API. - Architected RAG retrieval pipeline with ChromaDB vector store, multi-provider LLM orchestration (Claude, GPT-4, Gemini) with fallback routing, structured output validation, and token budget optimization — directly relevant to Glean's retrieval and intelligence infrastructure. - Built aeval evaluation platform with five eval types, adversarial safety testing, statistical significance testing, CI/CD regression detection, and automated safety gates — establishing the kind of measurement rigor that enterprise intelligence products require to earn customer trust. - Implemented OpenClaw multi-agent orchestration framework with gateway protocol, subagent delegation, and session switching — enabling coordinated AI agent workflows across multiple industry verticals. - Conducted enterprise-wide Service Language Assessment across nine languages, analyzing usage data and developer feedback to inform strategic investment decisions presented to the CTO — demonstrating comfort moving between ground-level data and executive-level strategy. **Closing** Glean's mission — making every employee AI-fluent and turning the superintelligent enterprise from concept into reality — is one I find genuinely compelling, and not abstractly so. I've spent the last several years building the infrastructure and evaluation systems that make AI trustworthy and useful in production, and I want to do that at the scale and stakes that Glean's enterprise customer base demands. I'd welcome the opportunity to discuss how my background maps to the specific product bets you're considering for Enterprise Intelligence. Thank you for your time and consideration. Sincerely, **O. Felix Amoruwa** famoruwa@berkeley.edu | 909-731-9011 | felixamoruwa.info