Personal Projects

---

Tiny-ml

A minimalistic machine learning library written purely in c++ from scratch.

• Tech stack: C++, CMake, Ninja, OpenMP, Shell
• Features:
  • Template based generic framework
  • Utilizes openMP framework for parallelization
  • Features modular library and scalable
  • in-built unit tests with performance benchmarking

tiny-ml repo

Terminus

A native CLI generative AI voice agent with chat mode.

• Tech stack: Python, OpenAI, Livekit, Typer, Argparse, Shell
• Features:
  • Features chat and voice mode with CLI flags
  • Agentic workflow execution for interacting with google calendar and email
  • in-built noise cancellation and latency optimization

terminus repo

Graphomerx

An extension of Graphomer framework for topological deep learning especially for small molecules

• Tech stack: Python, Rdkit, PyTorch, DeepChem, HuggingFace
• Features:
  • Built data pipeline for molecule using SMILES parsing & Rdkit graph construction
  • Extended graph transformer with topological embeddings
  • Benchmarked on QM9 dataset
  • Published model on HuggingFace