application_answers / art_EFLvIrj8vu0

Content

{
  "answers": [
    {
      "question": "Preferred First Name",
      "answer": "Felix"
    },
    {
      "question": "LinkedIn Profile Link",
      "answer": "https://www.linkedin.com/in/felixamoruwa"
    },
    {
      "question": "Github Link",
      "answer": "https://github.com/felixamoruwa"
    },
    {
      "question": "Personal Website About You",
      "answer": "https://felixamoruwa.info"
    },
    {
      "question": "Current Company",
      "answer": "Streamio AI / Fintellect AI"
    },
    {
      "question": "Current Title or Role",
      "answer": "Founder & CEO"
    },
    {
      "question": "Past Company 1",
      "answer": "Intuit"
    },
    {
      "question": "Past Company Title or Role",
      "answer": "Staff Product Manager \u2014 Developer Frameworks & Platform Infrastructure"
    },
    {
      "question": "Past Company 2",
      "answer": "Splunk Inc."
    },
    {
      "question": "Past Company Title or Role 2",
      "answer": "Senior Product Manager \u2014 Search Orchestration"
    },
    {
      "question": "(Optional) List 3 projects you're proud of.",
      "answer": "1. RL Workbench (2026): A 3-phase post-training platform covering the full RLHF/DPO pipeline \u2014 Reward Lab, Playground, and Arena \u2014 implementing 12 RL algorithms (PPO, GRPO, DAPO, DPO, SimPO, and others) with live SSE metric streaming and head-to-head benchmarking across TRL, VeRL, OpenRLHF, and NeMo RL in GPU-passthrough Docker containers.\n\n2. aeval \u2014 AI Model Evaluation Platform (2025\u20132026): A local-first evaluation platform with 5 core eval types, adversarial safety testing with refusal detection, and statistical rigor including bootstrap confidence intervals, Welch's t-test, and Cohen's d effect size. Built on FastAPI, TimescaleDB, Redis, Next.js, and Ollama with CI/CD regression detection and automated safety gates.\n\n3. BRAIN \u2014 Protein Structure Prediction ML Platform (UC Berkeley 2004; rewritten 2026): Originally hand-coded in C++ with custom BPTT, rewritten in PyTorch spanning 5 neural architectures (feedforward, GRU, Transformer, ESM-2, multi-task) from 413 to 8B parameters \u2014 a 19M-fold scale increase. The original research was accepted at NeurIPS 2014."
    },
    {
      "question": "Will you now or in the future require sponsorship for employment visa status in the United States?",
      "answer": "No \u2014 based on the candidate's profile and work history, no visa sponsorship is indicated as required."
    },
    {
      "question": "(Optional) Other notes",
      "answer": "I've spent the last two years building at the intersection of research and production AI \u2014 from hand-implementing 12 RL algorithms in a post-training workbench (GRPO, DPO, PPO, DAPO, and others) to publishing at NeurIPS on neural networks for protein structure prediction. At Intuit, I drove platform products that scaled to 675M+ engagements and reduced developer onboarding from weeks to minutes. I'm drawn to Thinking Machines Lab because the Research PM role maps directly to what I find most energizing: translating frontier research into well-scoped, executable programs while maintaining scientific rigor. I'm comfortable going deep on ML infrastructure (compute roadmaps, training pipelines, eval frameworks) and equally comfortable synthesizing that complexity for cross-functional stakeholders."
    },
    {
      "question": "Candidate Confidentiality Acknowledgment",
      "answer": "I agree"
    }
  ],
  "skipped": [
    "First Name",
    "Last Name",
    "Email",
    "Phone",
    "Resume/CV"
  ]
}