Skip to content
View dhruvkelawala's full-sized avatar
👨‍💻
Where are we?
👨‍💻
Where are we?
108 followers · 8 following

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx4Achievement: YOLOAchievement: QuickdrawAchievement: StarstruckAchievement: Arctic Code Vault Contributor

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx4Achievement: YOLOAchievement: QuickdrawAchievement: StarstruckAchievement: Arctic Code Vault Contributor

Block or report dhruvkelawala

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
dhruvkelawala/README.md

👋 Dhruv Kelawala

Production engineer building reliable systems around unreliable LLMs.

I ship AI agents, multi-agent orchestration, and anti-fabrication pipelines. Four years architecting high-stakes interfaces at Argent (2M+ downloads, 70% market share) — the same discipline that made wallet flows safe now makes LLM output trustworthy.

🤖 Shipped AI Products

Product What it does Tech
Visaroy Live AI visa prep assistant — 3-stage LLM pipeline with grounding validator, 21-country PDF engine, zero-data-retention Next.js · Vercel AI SDK · Gemini
FluperMario Production AI agent system — V8 isolate Code Mode, anti-fabrication evals (0 vs 4 baseline), 12 ADRs TypeScript · Flue · V8 Isolates

🛠️ Agent Infrastructure

Project What it does Tech
task-dispatch Agent orchestration — 9-state FSM, DAG dependencies, multi-agent QA pipeline TypeScript · SQLite
sumocode Agent UX layer — persistent memory, approval gating, from-scratch terminal renderer (983 tests) TypeScript · Yoga WASM
loopwatch Observability for AI coding agents — multi-agent session ingestion, convergence cockpit TypeScript · Tauri
skills Agent workflow tooling — cross-model code review, PR automation, process-as-skills Shell · AI Frameworks
hevy-cli Agent-native CLI — --json for every command, SKILL.md for agent consumption, WHOOP integration Go

🌟 Open Source & Scale

  • Core contributor to Argent X — 2M+ downloads, 70% Starknet market share
  • Built web3-starknet-react — React hooks & connectors used by thousands of developers (40★)
  • Core contributor to starknet.js — primary TypeScript SDK for Starknet

🚧 Now

  • Shipping Visaroy (live) and Forayy (LangGraph multi-agent pipeline)
  • Building agent infrastructure: orchestration, UX, observability
  • Based in London, open to AI product and agent infrastructure roles

📫 Reach Out

X / Twitter • LinkedIn • Website • Visaroy

"The LLM is unreliable. The system around it doesn't have to be."

Pinned Loading

  1. web3-starknet-react web3-starknet-react Public

    A Starknet React Provider and Connectors inspired by web3-react library

    TypeScript 40 11

  2. sumocode sumocode Public

    Personal Pi extension — OpenCode visual language, persistent memory, preattentive status signals

    TypeScript 2

  3. loopwatch loopwatch Public

    Loopwatch — a local, passive convergence layer for AI coding agents (Pi, Codex, Claude Code): watches sessions, recommends loops, flags drift. Design + v1 PRD.

    TypeScript 2

  4. hevy-cli hevy-cli Public

    CLI Tool and Agent Skill for the Hevy workout tracking API, written in Go

    Go

  5. skills skills Public

    Pi agent skills — clawpatch automated code review and more

    HTML

  6. task-dispatch task-dispatch Public

    Dispatch-native task management for OpenClaw ACP agents — Kanban, DAG dependencies, automated QA pipeline

    TypeScript 1