cover_letter / art_ACzmvCLWxOA

Dear Applied Intuition DroneStack Hiring Team, Applied Intuition is building the digital infrastructure for physical AI — the kind of foundational platform work that determines whether autonomous systems actually reach the field or remain perpetually in the lab. That mission is concrete and consequential, and the DroneStack challenge — transitioning from bespoke program deliveries to a scalable, multi-customer product line — is exactly the kind of 0-to-1 platform problem I have spent my career solving. When I built Intuit's ICE Self-Service platform from scratch, the core challenge was identical: take a collection of one-off integrations serving individual teams and turn them into a repeatable, developer-facing product that could scale across an entire enterprise. That experience, combined with my FAA certifications and hands-on AI systems work, is what draws me to this role. **Technical and Platform Foundation** My technical credibility runs deep and is directly relevant to the systems-level thinking DroneStack requires. At UC Berkeley I studied Computational Engineering Science, and my 2004 capstone project was a hand-coded neural network in C++ with custom backpropagation-through-time for protein structure prediction — work that was accepted at NeurIPS 2014. I have since rebuilt that system in PyTorch spanning 413 parameters to 8 billion, with MLflow experiment tracking, Optuna hyperparameter optimization, and Docker orchestration across six containers. More recently I built a full RL post-training workbench implementing 12 algorithms — PPO, GRPO, DAPO, REINFORCE, DPO, and others — with live metric streaming, GPU Docker passthrough, and head-to-head benchmarking across TRL, VeRL, OpenRLHF, and NeMo RL. These are not survey projects; they are production systems built to understand how real AI training infrastructure behaves under load. On the platform side, at Intuit I owned the ICE developer framework serving QuickBooks, TurboTax, Mint, Mailchimp, and Credit Karma. I scaled that platform from 6,000 to 50,000 transactions per second via rSocket migration supporting approximately 1.5 million concurrent connections at sub-25ms TP99, and grew engagements 275% year-over-year to 675 million in FY23. I also led 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 kind of market and technical synthesis that maps directly to building a product thesis for an emerging autonomous systems platform. I hold an FAA Commercial Drone License (sUAS Remote Pilot Certificate) and an FAA Private Pilot License. I understand airspace, flight operations, and the regulatory environment that shapes how autonomous aircraft are deployed — not as a hobbyist, but as a certificated operator who has studied the systems and constraints that govern real flight. **Why DroneStack, Why Now** The transition from program-driven delivery to scalable product is one of the hardest inflection points any technical team faces. It requires someone who can hold two things simultaneously: deep enough operator empathy to know which abstractions are real versus convenient, and enough platform instinct to know which configurations, interfaces, and tooling investments will compound across customers. That is the work I did at Intuit with ICE, at Splunk with Search Orchestration across three microservice backlogs, and at Kaiser Permanente scaling Splunk Logging-as-a-Service to 200 internal enterprise customers processing 1.7 TB daily. The domain is new; the problem structure is familiar. **Role-Specific Connection** What excites me most about this role is the operator interface challenge — designing mission control, mission planning, and mission review tools that non-technical users can operate without engineering support in the field. I have shipped developer-facing SDKs and self-service platforms specifically to remove the need for engineering hand-holding, and I understand the design discipline required to make complex systems feel simple under operational pressure. The requirement to conduct primary research with military operators in the US and abroad, synthesize doctrine and workflow patterns, and translate those findings into product strategy is precisely the research motion I applied when conducting enterprise-wide developer surveys at Intuit and beta customer benchmark testing at Splunk. **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-prod and under 24 hours for production, while mitigating $1M+ in projected opex growth. This is the programs-to-products transition Applied Intuition is asking DroneStack to make. - **Platform scale:** Scaled ICE throughput from 6K to 50K TPS via rSocket migration supporting ~1.5M concurrent connections with sub-25ms TP99; grew to 675M+ engagements in FY23 across five major product lines. - **Enterprise Service Language Assessment:** Conducted cross-functional analysis across nine languages and multiple engineering organizations, synthesizing usage data and developer feedback into strategic recommendations presented to the CTO — the same research-to-strategy motion required for building a drone market thesis. - **AutoEval — Automated Visual Evaluation for Robot Model Training (2025):** Repurposed StreamIO's screen capture and multimodal AI pipeline to score robot model outputs (grasp poses, segmentation maps, bounding boxes) against natural-language rubrics, reducing evaluation cycles from 72 hours to approximately 4 minutes. Zero-integration architecture captures from any visualization tool via screen capture; Claude/GPT-4V performs spatial reasoning on prediction frames and generates structured PASS/FAIL reports with confidence scores. - **Splunk Search Orchestration:** Owned Search Service (Go microservices), Search Catalog (PostgreSQL metadata service), and SPL/SPL2 — delivered Scheduler Service end-to-end in approximately four months, enabling scheduled search capabilities demoed at .conf19. Led query performance optimization achieving up to 10x improvements for a Fortune 500 beta customer. - **OpenClaw Multi-Agent Orchestration (StreamIO):** Implemented multi-agent framework with gateway protocol, subagent delegation, profile management, and session switching — enabling coordinated AI agent workflows across multiple industry verticals. Relevant to the autonomous mission coordination architecture underlying swarm operations. - **FAA Certifications:** FAA Commercial Drone License (sUAS Remote Pilot) and FAA Private Pilot License — operational familiarity with airspace regulations, flight systems, and the constraints that shape autonomous aircraft deployment. **Closing** Applied Intuition's mission — bringing intelligence to every moving machine on the planet — is one of the few technology bets where the outcome genuinely matters at a civilizational scale. DroneStack is at the sharpest edge of that mission: real platforms, real operators, contested environments. I want to help build the product layer that makes autonomous swarm capability repeatable, scalable, and trustworthy across customers and theaters. I would welcome the opportunity to discuss how my platform background, AI systems experience, and aviation credentials can contribute to that work. Respectfully, **O. Felix Amoruwa** famoruwa@berkeley.edu | 909-731-9011 | felixamoruwa.info