Skip to content
View Sandy-1711's full-sized avatar

Highlights

  • Pro

Block or report Sandy-1711

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.

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
Sandy-1711/README.md

Hi, I'm Sandeep πŸ‘‹

I work on AI agent infrastructure β€” the runtime layer that makes agents dependable: memory and semantic recall, streaming, tool orchestration, and background execution. I'm an active contributor to Mastra (25k+ ⭐) with 14 merged PRs across core, ai-sdk, and Studio. Previously: AI Engineer at AiRA (agent orchestration, memory, RAG) and AI Research Intern at IIT Kharagpur (multimodal LLM fine-tuning β€” 75% accuracy on circuit-generation tasks where base models scored ~0%).

πŸ“ Bengaluru, India Β· πŸ’Ό LinkedIn Β· πŸ“« sandy1711003@gmail.com Β· 🌐 Portfolio

πŸ”§ Highlight contributions

Explained so anyone can follow β€” each row is what I did and why it matters, in plain English.

Contribution What it means
The agent stopped forgetting Β· mastra #17818 AI agents keep a long-term memory, but on one type of incoming message they weren't checking it before replying β€” so they'd answer as if they'd forgotten past context. Now they look it up on every message.
No more double-typed replies Β· mastra #18191 When you open an answer that's already being generated and watch it stream in, some of the text was arriving twice. Fixed the duplicate delivery.
Background tools actually run in the background Β· mastra #18454 You can mark a tool to run in the background so the agent doesn't sit and wait for it β€” but that setting was being thrown away. Now it's saved and respected.
Your agent settings actually apply Β· mastra #18592 Tweaks you made to an agent in the visual editor weren't being used when you chatted with it through the API. Now they are.
Saving no longer wipes your work Β· mastra #18015 / #18404 Saving an agent in the editor was erasing its instructions and tools. Fixed two of those data-loss bugs.
Servers don't crash on simultaneous startup Β· mastra #18252 When two servers booted at the exact same moment, they could collide setting up the same message channel and fail. Fixed the race so startup is safe.
Added BigBlueButton video calls to cal.com Β· cal.com #29355 A whole new video-meeting option: booking a meeting can now auto-create a BigBlueButton room. Full integration β€” 912 lines of code, 32 tests.
Builds work even with a space in the folder path Β· n8n #20254 The project wouldn't build if its folder path had a space in it (common on Windows, like My Documents). Fixed so anyone can build it.

Full Mastra activity β†’ all PRs by me

πŸš€ Projects

  • Samagra β€” campus ERP platform I co-founded under Indigle: React Native + Node.js/Express + MongoDB, 10,000+ active users, 150k+ monthly API requests on a zero-cost serverless backend.
  • EmailPartner β€” turns incoming Gmail into live AI cards (headline, summary, illustration, voice narration): Gmail watch β†’ Pub/Sub push β†’ durable worker queue β†’ Gemini.
  • Web-Aware RAG Engine β€” async web ingestion and semantic search with load-balanced FastAPI replicas and a Next.js 15 frontend.
  • IntelliCommerce β€” intent-routing e-commerce chatbot across microservices with Pinecone-backed RAG.

πŸ›  Stack

Python TypeScript FastAPI Node.js Next.js React Native MongoDB PostgreSQL Redis Docker Mastra LangChain Langfuse promptfoo

Pinned Loading

  1. whoami whoami Public

    TypeScript 1

  2. mastra-ai/mastra mastra-ai/mastra Public

    Mastra is the modern TypeScript framework for AI-powered applications and agents.

    TypeScript 27k 2.6k

  3. EmailPartner EmailPartner Public

    Turns incoming Gmail into live AI cards β€” headline, summary, illustration, and voice narration per email. FastAPI + Pub/Sub push + durable worker queue + Gemini.

    Python 1

  4. Web-Aware-Rag-Engine Web-Aware-Rag-Engine Public

    Production-style RAG engine: async web ingestion, semantic search, and AI answers β€” load-balanced FastAPI replicas, Next.js 15 frontend.

    TypeScript 1

  5. IntelliCommerce IntelliCommerce Public

    An ecommerce chatbot capable of deciding intent based on the user query and redirect to different microservices.

    Python 1