Rohith K Bobby

India

Machine learning & backend developer, bug magnet, and occasional genius.

My Philosophy

Work should make sense—to me, to the machine, and to the person using it. Most good work starts by asking the right questions, not pretending to have all the answers. It's about finding those happy little solutions that stand strong, even when the world gets a little messy.

Core Toolkit

Python
PyTorch / TensorFlow
NLP
Computer Vision
Go / Rust
Docker
Git
Linux
Web Frameworks (FastAPI/Django)
System Design / Algorithms

Experience & Education

Machine Learning Intern

03/2023 – 12/2023

IHRD/SBCID Kerala

  • Automated newspaper analysis (OCR, Summarization), slashing processing time by 75%.
  • Enhanced case identification accuracy by 20% through NLP-driven keyword extraction.
  • Engineered an AI translation pipeline boosting Indic language accuracy by 25% (BLEU), empowering multilingual communication for >20 officials.

Research Intern

03/2022 – 03/2023

Aris4D

  • Pioneered a deep learning model for petri dish analysis, achieving 95% accuracy in predicting antibiotic effectiveness.
  • Reduced antibiotic selection time by 30% in pilot studies.
  • Cut model training time by 12% using computer vision for feature extraction.

Bachelor of Technology, Computer Science

11/2022 – Present

College of Engineering Kallooppara

Building theoretical and practical foundations in computer science and engineering.

Projects

goKyber

A High-Performance and Practical Post-Quantum Encryption Framework implemented in Go.

GoCryptographyPost-QuantumKyberSecurity
Abstract representation of goKyber

Malang

Malang is... well, let's just say it's a unique programming language. Born out of sheer defiance and an unwavering commitment to proving a point, Malang is my way of showing that I did pay attention in my compiler design course—whether my professor appreciates the outcome or not.

Compiler DesignProgramming LanguageGoHumor?
Abstract representation of Malang Language

Brain Tumor Segmentation & Classification

  • Leveraged ResU-Net architecture to achieve ~90% segmentation & classification accuracy, surpassing conventional CNNs by 15%.
  • Significantly reduced false positives (10%), enhancing potential diagnostic precision.
Computer VisionDeep LearningPyTorchMedical AI
Abstract representation of brain tumor segmentation

Image Search with Words

  • Developed a natural language interface for intuitive image library search.
  • Optimized embedding and search algorithms for near-instantaneous results.
NLPComputer VisionSearchPython
Abstract representation of image search

Diffusion MNIST

  • Implemented a Denoising Diffusion Probabilistic Model (DDPM) from foundational principles.
  • Successfully generated coherent MNIST digits purely from random noise, exploring controlled generative AI.
Generative AIDiffusion ModelsPyTorchCV
Abstract representation of noise to digit generation

Get In Touch

Open to interesting projects and collaborations. Feel free to reach out.

rohithkbobby04@gmail.com Rohith04MVK rohitk.