Kush

Kush Agarwal is an AI & ML engineer and open-source contributor — building from first principles in PyTorch and NumPy, and shipping LLM-powered tools that scale for real users.

Get in touch
AI & Machine Learning

Over the past year I have contributed to leading scientific libraries — merging a documentation PR into Kornia after navigating its strict CI/CD and review workflows, and earning recognition as an official contributor to the Gammapy v2.1 release, where I delivered a ~500x speedup on irregular geometry cropping through dynamic tensor slicing in NumPy. Beyond research libraries I ship full-stack and LLM-powered products — from agentic support systems with real tool-calling to Chalkie, a RAG-based app that synchronizes whiteboard visuals, voice narration, and a laser pointer to explain complex concepts. I study B.Tech Computer Science with an AI & ML specialization at GITAM, Visakhapatnam.

My open-source canvas.
Open Source.01
  • Merged a documentation PR into Kornia
  • Official contributor — Gammapy v2.1
  • ~500x speedup via dynamic tensor slicing
  • Navigated strict CI/CD & review flows
Learn more
Projects.02
  • GitHub AI Agent — LangGraph + MCP agentic workflow assistant
  • Agentic support system — live on HF Spaces
  • Word-embedding model in pure PyTorch
  • Chalkie — RAG-powered visual explanations
Explore projects
Skills.03
  • Languages — Python, C, C++, TypeScript, SQL
  • PyTorch, NumPy, Pandas, Flask, spaCy (NLP)
  • LangChain, LangGraph — agentic systems
  • Git, GitHub, CI/CD & system design
Learn more
Proof of work

Achievements & selected work.

Open-source contributions to scientific libraries and shipped products — each links straight to the PR, repo or live demo.