{
  "name": "Aryan Malik",
  "canonical_name": "aryan",
  "headline": "Looking for full-time work in San Francisco, January to May 2027: quant research and engineering, AI agents, RL environments and evals, market infrastructure.",
  "description": "Aryan Malik is graduating from Rutgers in December 2026 (math, computer science, economics) and is looking for full-time work in San Francisco from January to May 2027. He builds quantitative models, market systems, data infrastructure, and personal AI.",
  "updated": "2026-10-25",
  "knowledge": [
    {
      "id": "identity",
      "title": "About Aryan",
      "url": "/graph",
      "tags": ["about", "background", "quant", "engineer", "mathematics", "computer science", "economics", "Rutgers"],
      "text": "Aryan Malik studies mathematics, computer science, and economics at Rutgers University. He wants to become a quant engineer who can own the entire loop across research, development, and trading. His recurring interests include mathematics, coding, trading, optimization, football, health, and cars. He is deliberately trying to become exceptional and is making parts of that process visible."
    },
    {
      "id": "current-work",
      "title": "What Aryan is working on now",
      "url": "/work",
      "tags": ["now", "current", "working", "research", "projects", "normal", "period", "football", "Greeks", "neural marked point process", "RL trading swarm", "simulation", "reinforcement learning"],
      "text": "Aryan currently highlights three pursuits on /work: normal periods, prediction markets on himself, and his latest build. First, he is exploring whether irrational Kontsevich–Zagier periods are normal numbers, with pi as the motivating case; pi's normality is an open problem, so this is a research direction rather than a claimed result. Second, he runs play-money prediction markets about himself on this site (steps, sleep, Osmo usage), settled against public data. Third, his latest build is RL Trading Swarm (https://github.com/CANTSOAR/rl_trading_swarm): a closed simulated economy with stocks, options, bonds, a Taylor-rule central bank and a treasury, where an evolving population of neural-network traders lives, dies, splits and mutates. On 8 seeds × 20 simulated years, 14–16 of 17 stylized-fact checks pass on every seed with zero accounting violations; the realistic dynamics come from designed fundamentalist and sentiment traders, not emergent behavior. Paired policy experiments show a two-year +500bp rate peg drops the index about 30% (p = 0.002), and Tobin taxes don't reduce turnover. A PPO agent beats every baseline on 48 held-out seeds; with its operating costs switched off it earns +7.1% a year at a Sharpe of 1.7, and with normal costs every policy loses money. Writeup: https://aryanmalik.com/writing/rl-trading-swarm. A recent project is football greeks (https://github.com/CANTSOAR/football_greeks): a neural marked point process that simulates NFL games snap by snap to price team- and player-yardage contracts and their Greeks. On Kalshi's thin team-total books a pre-registered paper rule made about +$0.12 per contract over 49 games on minute candles; on 16 games recorded live with real quotes it made +$0.05 with a one-minute fill delay, too few games to confirm. On liquid player-yardage markets the market is as good as or better than the model, and no rule made money. Writeup: https://aryanmalik.com/writing/football-greeks. Every project writeup is at https://aryanmalik.com/writing."
    },
    {
      "id": "osmo",
      "title": "Osmo",
      "url": "/osmo",
      "tags": ["osmo", "agent", "AI", "memory", "tools", "system", "assistant"],
      "text": "Osmo is Aryan's personal agent and an extension of how he organizes his life. The full private system combines model routing, structured memory, tool servers, evaluation, automation, and several interfaces. Public Osmo is deliberately smaller: it answers only from an approved public knowledge base, cites sources, admits when it does not know something, and cannot reach Aryan's private memory, messages, health records, or accounts."
    },
    {
      "id": "markets",
      "title": "The Aryan Exchange",
      "url": "/markets",
      "tags": ["market", "prediction", "exchange", "health", "steps", "heart rate", "paper money", "accountability"],
      "text": "The Aryan Exchange is a work in progress: prediction markets about Aryan's real behavior, including steps, workouts, sleep, resting-heart-rate aggregates, and Osmo usage. It begins with non-redeemable paper money. Aryan does not trade in markets about himself. The current preview publishes exact oracle values and transparent historical reference odds; these are not presented as traded prices. The long-term design calls for shared public order books, explicit settlement receipts, downloadable history, and clear disclosure that Aryan is a reflexive underlying who may respond to market expectations."
    },
    {
      "id": "daily-loop",
      "title": "Daily Loop",
      "url": "/play",
      "tags": ["puzzle", "game", "loop", "leaderboard", "optimization", "daily"],
      "text": "Daily Loop is the small daily puzzle on Aryan's site, at /play. A visitor creates a simple username-and-password account or logs in, explicitly starts a visible timer, connects every point exactly once, and returns to the start. Score equals elapsed seconds multiplied by route distance, and the lowest score wins. Usernames are globally unique and each account receives one attempt per civil day. The board changes daily and the server proves the shortest possible distance exactly after submission."
    },
    {
      "id": "personal",
      "title": "Personal interests and stories",
      "url": "/graph",
      "tags": ["cars", "Miata", "football", "running", "Quebec", "friend", "memorable"],
      "text": "Aryan really likes cars and once rebuilt a Miata engine before selling the car. Friends describe him as very interested in finance and predictive models and as someone who likes fixing cars. One friend remembers running with Aryan in the severe Quebec cold around New Year's. Aryan also follows football and fantasy sports, enjoys mathematics and optimization, and likes telling good stories."
    },
    {
      "id": "public-signals",
      "title": "Public signals and privacy",
      "url": "/markets",
      "tags": ["signals", "data", "tokens", "spend", "health", "privacy", "telemetry"],
      "text": "Aryan wants accountability through visibility. The site publishes exact numeric Osmo usage totals and hourly and daily history, including input tokens, output tokens, cost, and turns. Its Google Health oracle publishes available hourly steps plus day-level steps, distance, active-zone minutes, sleep summaries, resting heart rate, and weight history. Current values and 24-hour or 14-day histories appear directly beside the markets at /markets; /agents lists every machine-readable interface. The machine-readable feed is /api/public/metrics.json. Raw prompts, messages, credentials, stable account identifiers, and other people's data are not betting values and are never published."
    },
    {
      "id": "hospitalgym",
      "title": "HospitalGym",
      "url": "/hire",
      "tags": ["HospitalGym", "hospital", "digital twin", "simulator", "simulation", "calibration", "reinforcement learning", "RL environments", "optimization", "health hack", "hackathon", "healthcare", "recent", "latest", "new project", "PPO"],
      "text": "HospitalGym is a hospital operations twin Aryan built at RU Health Hack in October 2026: build the twin from public data, calibrate a baseline, simulate, optimize. The first hospital is a model of Robert Wood Johnson University Hospital in New Brunswick, built from CMS, NHAMCS, and BLS data; it is not affiliated with the hospital and every result is simulated. Of 129 hospital-specific attributes, 89 are measured, 6 come from literature, and 34 are assumed. The calibrated baseline reproduces published figures on held-out seeds (median ED time for discharged patients 203 minutes against 206 published; 0.91% left without being seen against 1%). A batched PyTorch simulator runs thousands of hospitals in parallel for policy search. On held-out seeds, a tuned rule-based plan cut mean door-to-provider wait by 17.0% (95% CI 9.1 to 26.2) at 4.0% lower labor cost with every constraint met, holding in 20 of 24 calibrations that public data cannot tell apart; it misses a constraint under a flu surge. The PPO agents have not beaten the tuned plans. Writeup: https://aryanmalik.com/writing/hospitalgym · Source: https://github.com/CANTSOAR/healthhacks26/tree/aryan. Recorded demo: https://cantsoar.github.io/healthhacks26/"
    },
    {
      "id": "poscere",
      "title": "Poscere",
      "url": "/hire",
      "tags": ["Poscere", "health hack", "hackathon", "healthcare", "virtual patient", "simulator", "medical education", "recent", "latest", "new project", "MCP", "evals"],
      "text": "Poscere is a telehealth virtual-patient simulator for health-professions education that Aryan built with one teammate at RU Health Hack in October 2026. A student interviews an AI patient while a chart fills in, then gets rubric feedback tied to quotes from the transcript. Code, not the model, decides which facts the patient may release, and a second model checks every reply against the case. It has 63 learner roles and 81 seed patients defined as data (none reviewed by a clinician), Care Journeys that take one patient through several professions with a shared, traceable patient state, and an MCP endpoint for searching stored visits. Writeup: https://aryanmalik.com/writing/poscere · Source: https://github.com/CANTSOAR/lowhangingfruit26"
    },
    {
      "id": "looking-for",
      "title": "What Aryan is looking for",
      "url": "/hire",
      "tags": ["job", "jobs", "hire", "hiring", "looking", "looking for", "available", "availability", "when", "where", "role", "roles", "full-time", "internship", "co-op", "contract", "founder associate", "san francisco", "SF", "january", "may", "2027", "recruiter", "startup", "interested", "work with", "talk", "why", "quant research", "AI agents", "RL environments", "evals", "reinforcement learning", "market infrastructure"],
      "text": "Aryan is looking for full-time work in San Francisco from January to May 2027, and is open to longer. He is graduating from Rutgers in December 2026. He is interested in quant research and engineering, AI agents, RL environments and evals, and market infrastructure, and is open to internship, co-op, contract, founder-associate, or full-time arrangements. He will be in San Francisco Nov 1–6, 2026 and is happy to meet in person. Good reasons to talk to him if you build RL environments or evals: he built a Rust exchange and turned it into KnightGym, a Gymnasium/PettingZoo environment, and audited its reward for hacking. Contact aryan@aryanmalik.com or book time at /hire."
    },
    {
      "id": "education",
      "title": "Education and graduation",
      "url": "/hire",
      "tags": ["graduate", "graduation", "graduating", "graduates", "when", "degree", "rutgers", "december", "2026", "education", "major", "school", "honors college", "mathematics", "computer science", "economics"],
      "text": "Aryan graduates from Rutgers University–New Brunswick (Honors College) in December 2026 with a B.S. in Mathematics, Computer Science, and Economics."
    },
    {
      "id": "resume",
      "title": "Aryan's resume",
      "url": "/resume.pdf",
      "tags": ["resume", "cv", "curriculum vitae", "pdf", "download", "experience", "background"],
      "text": "Yes, Aryan has a one-page resume: https://aryanmalik.com/resume.pdf. It covers LuminArx Capital Management, his independent quant research, Gradient Trading, the New Jersey Big Data Alliance, and his projects (Osmo, football Greeks, Trade the Knight and KnightGym)."
    },
    {
      "id": "contact",
      "title": "Working with Aryan",
      "url": "/graph",
      "tags": ["contact", "email", "reach", "hire", "collaborate", "role", "work", "quant engineer", "linkedin", "github", "meet", "book", "schedule"],
      "text": "Contact: email aryan@aryanmalik.com, LinkedIn https://www.linkedin.com/in/aryan-malik-xd/, GitHub github.com/cantsoar. Resume: https://aryanmalik.com/resume.pdf. To book time, use https://aryanmalik.com/hire (or ask Osmo to check his availability and book). Aryan wants to meet people working on difficult problems in quantitative research, market and financial systems, data infrastructure, AI agents, RL environments, and adjacent fields. The strongest introduction points to a concrete question, dataset, model, or system worth building."
    }
  ]
}