Skip to content
View mac146's full-sized avatar
  • 14:37 (UTC +05:30)

Block or report mac146

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. 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
mac146/README.md

Mayank

Backend engineer with a thing for embedded systems and developer tooling.

I work across the full depth of the stack — TypeScript APIs, Python backends, and C++ firmware when the job calls for talking directly to hardware. I'm currently deep into AI tooling, specifically MCP (Model Context Protocol) and RAG systems.


Projects

Adaptive-RAG Python FastAPI Qdrant BM25

A structure-aware RAG system that actually thinks before it retrieves. It parses PDFs, DOCX, and Markdown, detects document structure and sections, then picks a retrieval strategy based on both the document layout and the question type. Retrieval is hybrid — Qdrant for vector search, BM25 for keyword — followed by cross-encoder reranking before the LLM ever sees the context. Most RAG implementations skip all of this and just do nearest-neighbour on the whole doc. This one doesn't.

IssueWatcher TypeScript MCP Node.js

Monitors all your starred GitHub repos for new issues, runs each one through an LLM, and sends you an email with a plain-english summary and a suggested fix. Built on MCP so you can swap the underlying model or add new sources (JIRA, HackerNews, Reddit) without touching the core logic. One of those tools I built because I actually wanted it.

FastPay Python FastAPI MongoDB

REST API for a digital wallet and transaction system. Users, wallets, cross-user transactions — full CRUD with auto-generated Swagger docs at /docs. Uses PyMongo directly, no ORM. Proper .gitignore, requirements.txt, and project structure from the start.

soccer-bot C++ Arduino PlatformIO

Autonomous robot built on Arduino Nano (ATmega328), written in C++ with PlatformIO. Proper layout — src/, include/, lib/, test/. Different problem space from web work entirely: no stack traces, no hot reload, just hardware doing what the code says or not.


Stack

Languages   TypeScript · Python · C++ · JavaScript . C
Backend     FastAPI · Node.js · MongoDB · PyMongo
RAG/AI      Qdrant · BM25 · cross-encoder reranking · LLM APIs · MCP
Frontend    React · CSS
Embedded    Arduino Nano · ATmega328 · PlatformIO
Tooling     GitHub REST API · Nodemailer · Swagger UI · uvicorn

Stats


27 repos  ·  India, UTC +05:30  ·  open to collabs and interesting problems

Pinned Loading

  1. fastpay fastpay Public

    Python

  2. IssueWatcher IssueWatcher Public

    Ranks new issues in the GitHub repos you star and depend on, flags the ones you could pick up, and emails a digest. AI summaries optional; runs free on GitHub Actions.

    TypeScript 2

  3. soccer-bot soccer-bot Public

    C++

  4. adaptive-rag adaptive-rag Public

    TypeScript