Skip to content
View Kaynup's full-sized avatar
🏠
Working from home
🏠
Working from home

Block or report Kaynup

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. 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
Kaynup/README.md
Typing SVG

About Me

Electrical Engineer with CS Specialization. Building Full Stack Systems and applications along with data-oriented projects, AI applications, Open source contributions and basically everything that tires me out...

Note: Some of these projects are currently private or in active development until their official release

Kaggle HyperShort Orpheus cythml

Earth and Connections

Open Source Contributions

I actively contribute to popular open source repositories. Here are some of my recent contributions:

  • scikit-learn/scikit-learn: PR #34817 - Optimized confusion_matrix_at_thresholds with integer cumulative sum for unweighted inputs.
  • scikit-learn/scikit-learn: PR #34789 - Preserved pandas DataFrame inputs and feature names across RFE and RFECV.
  • explosion/projects (spaCy): PR #211 Replaced OS-specific bash commands with cross-platform inline Python.

Featured Projects

Production-grade URL shortener: FastAPI + ReactJS, JWT auth with sessionStorage, DynamoDB, bijective short-code generation, and 302 redirection. Deployed on AWS EC2 (t3.micro) with IAM role auth, elastic IP, and Caddy SSL.

(Private Repository)

Full-stack deployment of sentiment classifiers in Scikit-learn and JAX frameworks. Inferences handled via FastAPI on a Railway backend, connected to a vanilla JS frontend on Vercel.

Optimized machine learning algorithms and tools implemented with Cython for high performance and reduced Python overhead.

(Private Repository)

Structured a robust research workspace focusing on multimodal product price prediction from catalog text descriptions and images. Benchmarked TF-IDF + LightGBM to BERT fine-tuning regressors.

Implemented a tabular Q-Learning agent for Tic-Tac-Toe. Fully reproducible training using configurable hyperparameters and random seeds via config.yaml.

Retrieval-Augmented Generation (RAG) system with persistent vector storage (ChromaDB), multi-provider LLM orchestration (LiteLLM), and strict source citations. Built with Flask backend

(Private Repository, public on beta release)

Technical Stack

Category Technologies
Languages Python Java JavaScript ReactJS Go Cython
Backend FastAPI Flask
Deployment Docker AWS Vercel Railway
Machine Learning & AI Scikit-learn PyTorch llama.cpp Ollama
Data Eng & Storage Apache Spark PySpark PostgreSQL MySQL DynamoDB

GitHub Stats

GitHub Stats Top Languages

Pinned Loading

  1. BIRD-CLEF-2026-Kaggle-Competition-Project BIRD-CLEF-2026-Kaggle-Competition-Project Public

    Jupyter Notebook

  2. Amazon-ML-Hackathon-2025-Research-Educative-Pipelines Amazon-ML-Hackathon-2025-Research-Educative-Pipelines Public

    Jupyter Notebook

  3. DLMMDD-Workshop-Synthetic-Image-Attribution-Kaggle-Competition DLMMDD-Workshop-Synthetic-Image-Attribution-Kaggle-Competition Public

    Jupyter Notebook

  4. Family-Support-Scheduler-Full-stack-App Family-Support-Scheduler-Full-stack-App Public

    Python

  5. Playground-Series-S6-E4-Kaggle-Competition Playground-Series-S6-E4-Kaggle-Competition Public

    Jupyter Notebook 1

  6. Binary-Text-Classification-with-Sklearn-Jax Binary-Text-Classification-with-Sklearn-Jax Public

    Jupyter Notebook