Sabyasachi Patra

Senior Go Engineer — Distributed Systems, P2P & AI Agent Infrastructure
Available for opportunities — remote, contract or full-time
Kolkata, India · Open to Remote (EU / US)
mail asabya@gmail.com
github github.com/asabya
linkedin linkedin.com/in/sab94

Senior Go engineer with 8+ years building production distributed systems, decentralized storage, and P2P networking infrastructure. Primary author of FairOS-dfs and contributor to Ethereum Swarm's Bee client. Now applying deep protocol-level systems expertise to AI agent infrastructure—building decentralized agent marketplaces, local memory systems, MCP servers, and autonomous coding workflows.

Experience

Datafund (EU) — Lead Engineer, Decentralized Storage Nov 2021 – Aug 2025
Go · Swarm · Ethereum · Remote
  • Primary author of FairOS-dfs — decentralized file system built on Swarm; 29 stars, 12 forks, Apache-2.0.
  • Architected pod-based storage, user management, ACL (Access Control Trie), group drives, and ENS-based authentication from the ground up.
  • Built HTTP REST API layer with OpenAPI docs, WASM compilation target, and FUSE client for desktop mounting.
  • Contributed upstream fixes to ethersphere/bee (Swarm's core Go client).
  • Implemented Web3 data assetization workflows and Git integration for decentralized repo storage.
Datahop Labs (UK) — Go Engineer, P2P / IPFS Apr 2021 – Mar 2023
Go · IPFS · libp2p · Contract / Remote
  • Primary contributor to datahop/ipfs-lite — lightweight IPFS node for resource-constrained environments.
  • Implemented peer discovery (mDNS, DHT), content routing, and data exchange protocols optimized for mobile and low-bandwidth networks.
SWRM Labs (US) — Software Engineer, Go Backend Nov 2019 – Mar 2021
Go · Distributed Systems · Remote
  • Built production backend services for distributed networking infrastructure with focus on reliability and fault tolerance.
  • Participated in system design for long-term architecture of decentralized platform components.
Math & Cody — Open Source Backend Consultant Dec 2018 – Nov 2019
Go · IPFS · Remote
  • Built the Go backend for IPFSdrop, a local P2P file-sharing application.
  • Contributed to open-source P2P projects and Go-based system design.
Adience — Web Developer Feb 2017 – Jul 2018
  • Built web applications across client projects; gained production experience in full-stack development and deployment.

AI & Agentic Development

Nightmanager — Spec-Driven Pi Agent Workflow github
  • Built a Pi extension/package that turns intent into specs, vertical TODOs, delegated subagent implementation, validated commits, and reviewable PRs.
  • Designed finder/oracle/librarian/worker/manager handoffs, lean prompt loading, audit artifacts, and unit/integration/e2e test coverage.
Picobrain — Local Semantic Memory for AI Agents github
  • Built a local MCP memory server in Go using SQLite, sqlite-vec, local embeddings, structured thought/claim storage, semantic search, and import/export.
  • Added graph memory primitives, typed relationships, neighbor/path queries, SpaCy sidecar health checks, Docker/Goreleaser releases, and Pi extension tools.
Betar — Decentralized AI Agent Marketplace github
  • Built a fully decentralized P2P marketplace where autonomous AI agents discover, negotiate, and transact via libp2p + x402 micropayments (USDC on Base).
  • Agents register as ERC-721 NFTs, listings replicate via CRDT over GossipSub, and execution happens over direct libp2p streams—no central server.
  • Integrated Google ADK/Gemini, TUI, CLI, SDK, React frontend, EIP-712 USDC authorizations, and TCP/QUIC networking with DHT bootstrap.
Cog — Multi-LLM AI Assistant github
  • Built a multi-provider AI assistant (Claude, GPT-4, Grok) with full MCP tool integration, automatic tool discovery and chaining, context window management, and conversation persistence. Python/FastAPI with CLI and REST API interfaces.
Gemini OpenAI Proxy — OpenAI-Compatible Gemini API github
  • Built an async FastAPI proxy that exposes Google Gemini through OpenAI-compatible endpoints (/v1/chat/completions, /v1/models), enabling drop-in use with existing OpenAI SDKs. Browser cookie auth, streaming support, multi-model routing, Docker deployment. 5★.