Harsha Vardhan Vasa

AI/ML Engineer experienced in building machine learning systems end-to-end, from data preparation and feature engineering to model training, evaluation, and deployment-ready workflows.

Cybersecurity Essentials, Git for Beginners, Introduction to Cybersecurity, Introduction to IoT, Learning Linux Basics.

Mitigating Counterfeiting Using Blockchain Enabled Product Authentication

International Journal for Research in Applied Science & Engineering Technology (IJRASET).

Languages: Python, SQL, Bash

Frameworks: PyTorch, TensorFlow/Keras, Scikit-learn, FastAPI/Flask, Hugging Face Transformers 

Libraries: NumPy, Pandas, OpenCV, SciPy, Matplotlib, Seaborn, SHAPm MLflow

Concepts: Supervised Learning, Deep Learning, Computer Vision, Feature Engineering, Cross-Validation, Hyperparameter Tuning, Class Imbalance Handling, Regularization, Threshold Optimization, Model Calibration, Error Analysis, Ablation Studies, Explainable AI (XAI), Drift Monitoring, Metrics (Precision/Recall, F1, ROC-AUC)

Production-Style ML System: End-to-End Model Development, Tuning & Robust Evaluation

Jan 2025 – Apr 2025

• Built a full ML workflow from data preprocessing and feature engineering to model training, hyperparameter optimization, and final selection using cross-validation.
• Designed robust evaluation using precision/recall, ROC-AUC/F1, calibration analysis, and stress testing to validate performance across edge cases and potential drift scenarios.

ML Experimentation Platform: Reproducible Training, Model Comparison & Explainability

Jul 2024 – Nov 2024

• Developed a structured ML experimentation framework using MLflow to track parameters, metrics, datasets, and artifacts across multiple training runs for reproducible results.
• Added explainability using feature importance and SHAP-style interpretation to identify key prediction drivers and improve debugging and model trust.

Detection of Potato Leaf Diseases Using Deep Learning

May 2022 – Aug 2022

• Developed a deep learning-based computer vision system achieving 98% accuracy.
• Optimized disease identification workflows and designed a user interface to improve diagnostic efficiency.

Mitigating Counterfeiting Using Blockchain Enabled Product Authentication

International Journal for Research in Applied Science & Engineering Technology (IJRASET).

GIT for Beginners, Learning Linux Basics Course & Labs, AWS Academy Graduate - AWS Academy Cloud Foundations, CCNA: Introduction to Networks, Cybersecurity Essentials, Introduction to Cybersecurity, Introduction to IoT, Salesforce Developer Virtual Internship