Hemanth S Banur

Machine Learning & MLOps Enthusiast

Profile Picture

Summary

Passionate Machine Learning and MLOps enthusiast with a strong foundation in Computer Science. Experienced in developing ML models, implementing MLOps practices, and creating web applications.

Skills

Languages:

C/C++, C#, Python, JavaScript, TypeScript, MATLAB, HTML, CSS

Libraries:

Pandas, Numpy, Scikit-Learn, OpenCV, PyTorch, TensorFlow, Matplotlib, Seaborn, NLTK

Frameworks/Tools:

Docker, React.js, Node.js, MongoDB, Git, Unity, Azure, mlflow, ZenML, Flask, Dagshub, Firebase, Django, Flutter, Linux, Transformers, LLMs

Education

PES University

Nov 2022 - Present

Bangalore, India

B.Tech in Computer Science and Engineering

Experience

IEEE CS, PES University

ML Research Intern

Jul 2024 - Aug 2024, Remote

Bangalore

  • Architected and implemented a novel hybrid GAN-VAE system for artistic style transfer, combining an encoder for feature extraction with advanced generative networks, achieving 40% improved style preservation compared to baseline models
  • Led comprehensive comparative analysis of multiple GAN architectures (Cycle GAN, Attention GAN), optimizing the final three-component architecture through systematic experimentation and achieving state-of-the-art results in artistic transformation quality

Toyota Kirloskar Motor

Data Science Intern

Jun 2024 - Jul 2024, On-Site

Bangalore, India

  • Developed and deployed an enterprise-level automated sentiment analysis system using NLTK and Python, revolutionizing review processing from 7 hours to 10 seconds, while enabling seamless access for 100+ employees across company LAN infrastructure
  • Engineered a real-time data analysis pipeline with 99% efficiency improvement (4 hours to 5 seconds), incorporating robust DevOps practices for deployment and maintaining 100% data accuracy across multiple analytics solutions

CDSAML, PES University

Research Intern

Jun 2024 - Jul 2024, On-Site

Bangalore

  • Developed a breakthrough Kannada-to-English speech translation pipeline surpassing Google’s Long Model benchmarks, implementing a comprehensive seven-stage system for real-time processing
  • Engineered state-of-the-art audio processing modules incorporating advanced ML techniques (denoising, enhancement, segmentation, transcription), achieving superior accuracy in cross-lingual translation while maintaining real-time performance

Projects

DevForge

  • Engineered an AI-powered development environment integrating LLMs with containerized runtimes, automating code generation, execution, and error-correction to enhance developer productivity and code reliability
  • Developed a web-based IDE with seamless project management and interactive console access, orchestrating communication between Linux filesystem, cloud containers, and AI models for a streamlined end-to-end development workflow

LLMsDockerpythonReact.js

MLOps - Credit Card Fraud Detection

  • Architected and implemented a comprehensive end-to-end MLOps pipeline for credit card fraud detection, encompassing robust data ingestion, preprocessing, model training, and production deployment workflows
  • • Established a scalable MLOps infrastructure with version control, CI/CD, automated testing, and advanced experiment tracking/model management using DagsHub and MLflow, ensuring reproducibility and maintainability of machine learning systems

dagshubmlflowpython

Cropify

  • Developed a full-stack agricultural assistance platform integrating React.js frontend and Flask backend, leveraging ML/DL models to provide personalized crop and fertilizer recommendations based on environmental and location-based data
  • Implemented real-time plant disease detection using deep learning, with automated diagnosis and treatment recommendations, demonstrating the practical application of computer vision techniques to empower data-driven farming

scikit-learnReact.jspython

Activities & Achievements

Practical Approach to ML

Won 1st place in the hackathon organised by CIE PESU (Apr 2024)

Appex | Technical Lead

Organized 24-hr hackathons and ideathons for 200+ students and provided mentorship and resources.

Aura | Technical member

Organized "Crayion", a hackathon focused on image generation using advanced models such as DALL-E and MidJourney and served as mentor for the event.