Skip to content
View SatyamDave's full-sized avatar

Block or report SatyamDave

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
SatyamDave/README.md

Satyam Dave

I build agents that do real work, and the systems that keep them honest.

Now: founding engineer at a startup in stealth, owning product, engineering and strategy.

On the side: Real World, a simulation engine that seals a forecast about real events every morning and scores itself when the outcome lands. Founders use it to see how their customers react to a message before they ship it.


Agents and dev tools

  • parley: run several Claude Code agents on one repo; it plans a goal into a task graph, dispatches it and explains every change back to you
  • simulation-labs: a CI gate where computer-use agents attempt your signup and checkout on every deploy, and the merge is blocked when the flow breaks
  • index-one: a tamper-evident chain of authority for multi-agent actions, written in Rust: proof of who authorized what across companies
  • PHANTOM: autonomous black-box web pentesting with browser agents. Best Dev Tool, YC Hackathon 2026
  • llm-detect: identifies the model and provider behind an endpoint

Simulation and forecasting

  • real-world: message testing across 27 customer segments, backed by a public forecast track record
  • Regulars: runs thousands of simulated conversations, then checks them against what happened at a real dinner
  • brain-research: Against Synthetic Optimism, which measures and corrects sycophancy bias in LLM-simulated buying committees
  • clinical-trial-agents: agents predict clinical trial outcomes, checked against blinded results

Markets

  • quant-core: an open template for agentic trading, where the agent proposes trades and a deterministic Rust engine enforces the risk rules
  • fall_forward_labs: a research platform with event replay, point-in-time data and walk-forward validation
  • Fifty: one autonomous session on a Robinhood Agentic account through the Robinhood Trading MCP
  • bloombet: a Bloomberg-style terminal for sports-market inefficiencies

Products I've shipped and explored

Pluto (an AI assistant you reach by SMS and voice) · Vera (prompt to a full-stack app) · pigeon-ai · pom-pilot · FinalRoundAI · feedback-analyzer (Cloudflare Workers + D1)


TypeScript · Python · Rust · Go · Cloudflare Workers · Claude / LLM agents · Purdue

Pinned Loading

  1. fall_forward_labs fall_forward_labs Public

    Trading research platform: Rust event-replay engine with risk checks, point-in-time datasets and walk-forward validation.

    Python

  2. parley parley Public

    Run several Claude Code agents on one repo: plan a goal into a task graph, dispatch it, and explain every change back to you.

    TypeScript

  3. realworld realworld Public

    Message testing for consumer founders: simulate how your ideal customer reacts, with a public hashed forecast record. Apache-2.0.

    TeX

  4. Regulars Regulars Public

    Simulate the conversation, then check it against what happened at dinner: thousands of simulated trials, one real dinner a week.

    Python

  5. simulation-labs simulation-labs Public

    Behavioral CI gate: computer-use agents attempt your signup and checkout on every deploy and block the merge when the flow breaks.

    Python

  6. quant-core quant-core Public

    Open-source template for agentic trading: an AI agent proposes trades, a deterministic Rust engine enforces your risk rules.

    Python