Hi! I'm Ayush, a full-stack engineer and open-source contributor. I build web apps, AI integrations, and automated pipelines. Currently, I'm pursuing a B.Tech in Computer Science alongside a minor in AI & Data Science.

I like writing clean, readable code and figuring out how to scale systems. Most of my work involves TypeScript, Python, Next.js, and Supabase. Lately, I've been focused on building custom RAG architectures, database tools, and contributing to core web infrastructure.

Writing open-source software is one of my favorite ways to learn and give back. Here are some of my recent contributions:

• Wikimedia Foundation (MediaWiki Core)

  • Worked on UI performance patches and resolved front-end bugs for the Vector 2022 skin, which serves millions of users globally.
  • Debugged cross-browser issues and worked with senior maintainers using Gerrit, PHP, C++, and JavaScript.

• n8n Community Integrations

  • Created n8n-nodes-nvidia-nim to help developers run NVIDIA NIM models in their workflows.
  • Built custom nodes for Chat LLMs and Vector Embeddings, handling credential authentication and custom API request structures.

An AI-driven technical mock interview platform that simulates live coding and oral rounds.

• Dynamic Questions: Uses Gemini and NLP pipelines to ask candidates customized technical questions based on their domain.
• Feedback Engine: Evaluates candidate voice and text answers in real-time, outputting structured feedback on system design and coding concepts.

An open-source developer mentorship and CRM platform. It acts as a database and matchmaker for 185+ organizations across GSoC, LFX, and Outreachy.

• Smart Matching: I used a RAG pipeline with pgvector to match mentors and mentees based on skills and profile data.
• Architecture: Developed within a Turborepo monorepo architecture using Next.js and Supabase.

An enterprise energy auditing tool built to automate load calculations and report generation.

• AI Bill Parsing: Created a multi-modal pipeline using Gemini 1.5 Flash to extract consumption metrics from scanned MSEDCL utility bills with 99.8% accuracy.
• Automated Excel Audits: Wrote a formula-safe backend engine to programmatically generate ready-to-use engineering reports without breaking cell references.
• Stack: Next.js 15, Supabase Row-Level Security, Gemini API.

A real-time game analytics engine designed to predict and adapt gameplay difficulty dynamically.

• ETL Pipeline: Built a data pipeline using Apache Spark to process over 10,000 gameplay events.
• XGBoost Classifier: Used XGBoost to build predictive models that adjust question complexity on the fly.
• Stack: Python, Flask, Apache Spark, XGBoost.

A unified agritech service platform for rural accessibility.

• Logistics & Routes: Implemented Dijkstra's algorithm to calculate the most efficient last-mile delivery routes for local farmers.
• Crop Diagnosis: Built a FastAPI microservice that processes crop images to identify plant diseases using deep learning models.
• Stack: FastAPI, React, Python, Dijkstra's.