cover_letter / art_Y4lcI3qOiew

Dear Perplexity Hiring Team, Perplexity is doing something genuinely rare: turning the act of asking questions into a complete loop of learning, building, and acting — and with Computer, extending that loop into agentic action at scale. That mission resonates directly with the work I have spent the last several years doing: building AI-native platforms that transform information into decisions, from real-time market analysis agents to multi-agent orchestration frameworks to RL post-training workbenches. I am applying because the intersection of agentic AI, knowledge work, and enterprise productivity is exactly the space I have been building in — and I want to do it at the company defining the category. **Technical and AI Foundation** My AI/ML work is hands-on and spans the full stack. In 2026, I built a production RL post-training workbench covering the complete RLHF/DPO pipeline: a Reward Lab for designing and A/B testing reward functions across GSM8K, MATH, HumanEval, and UltraFeedback; a Playground running real TRL-powered GRPO and 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, DPO, SimPO, and others — with standardized throughput, memory, and convergence benchmarking. This is the kind of work that gives me a concrete mental model for evaluating nondeterministic model behavior and steering it toward high-value outcomes, which is precisely what Perplexity's JD calls out. I also 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. The stack — FastAPI orchestrator, TimescaleDB, Redis job queue, Next.js dashboard, Ollama — was designed for CI/CD integration with regression detection and automated safety gates. This is directly relevant to working with research teams to evaluate and steer models in production. My NeurIPS 2014 publication on artificial neural networks for protein secondary structure prediction, and the 2026 PyTorch rewrite of that original C++ BPTT system — scaling from 413 to 8 billion parameters — reflects a research foundation that goes back to the earliest days of deep learning applied to real scientific problems. **The Arc** From NeurIPS research to shipping production AI products to scaling developer platforms to 675M+ engagements at Intuit, my career has consistently been at the boundary between deep technical systems and the product decisions that make them useful to real people. The Perplexity PM role — working closely with engineering and design on Computer and search, with a mandate to envision new experiences and build data-driven flywheels — is the natural next step in that arc. **Why This Role** Computer represents the shift from AI as a research tool to AI as an action layer, and the enterprise productivity opportunity in that transition is enormous. My domain depth in financial services — I am a licensed real estate broker, a Topstep-funded trader, and the founder of Fintellect AI, a RAG-powered financial education and investing platform — gives me a concrete lens for identifying where agentic AI creates the most leverage in knowledge work. I am particularly interested in how Computer can accelerate decision workflows in investing, real estate, and financial analysis, industries where the gap between information and action is both costly and well-defined. I want to help Perplexity define those enterprise use cases and build the product experiences that make them real. **Selected Prior Experience** - Delivered ICE Self-Service platform at Intuit (DevPortal, GitOps config, 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 — a product-led growth outcome driven by removing friction at the adoption funnel. - 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. - Built OpenClaw multi-agent orchestration framework with gateway protocol, subagent delegation, profile management, and session switching — enabling coordinated AI agent workflows across real estate, insurance, health/dental, and financial markets industries. - Architected RAG retrieval pipeline for Fintellect AI with ChromaDB vector store, multi-provider LLM orchestration (Claude, GPT-4, Gemini) with fallback routing, structured output validation, and token budget optimization. - Worked closely with telemetry and usage data (SQL, BigQuery) to prioritize developer pain points across approximately 20 mobile apps and 30+ product SKUs at Intuit; built Asterias, a declarative asset lifecycle management platform with GraphQL API. - Led query performance optimization initiative at Splunk for a beta Fortune 500 customer, building a mirrored Enterprise topology for benchmark testing and achieving up to 10x performance improvements in Splunk Cloud Services search. - Designed repeatable RICE-based prioritization framework across three microservice backlogs at Splunk, balancing internal partner, third-party developer, and Fortune 500 customer requirements — the kind of conviction-under-uncertainty prioritization that small, high-velocity teams require. **Closing** Perplexity's mission — powering curiosity as a continuous cycle of learning, building, and integrating — is not a tagline I am adopting for this application. It describes how I have actually worked: from hand-coding BPTT in C++ at Berkeley in 2004, to publishing at NeurIPS, to building production AI platforms across fintech, real estate, and developer infrastructure. I want to bring that same compounding curiosity to the team defining what agentic AI looks like for the next generation of knowledge workers. I would welcome the opportunity to discuss how my background maps to the specific opportunities you are pursuing with Computer and search. Sincerely, **O. Felix Amoruwa** famoruwa@berkeley.edu | 909-731-9011 | felixamoruwa.info