Ragiri Himadeep
AI/ML Engineer & Data Scientist
Self-taught AI Engineer and Data Scientist building production-ready Generative AI, Machine Learning, & Deep Learning systems from scratch.
I’m a BBA Business Analytics graduate with a relentless passion for AI, Machine Learning, and Data Science. Self-taught from the ground up, I’ve transitioned from business analytics to building production-ready AI and Data Science systems that solve real-world problems. With 3+ years of self-study and 30+ projects across ML, DL, NLP, Generative AI, and MLOps, I specialize in end-to-end ML pipelines, cloud deployments, cutting-edge NLP, deep learning, and data-driven insights through advanced analytics. My goal is to create scalable AI and Data Science systems that drive business impact and innovation.
About Me
I’m Ragiri Himadeep, a self-taught AI/ML Engineer and Data Scientist with a BBA – Business Analytics degree. Despite not holding a traditional Computer Science degree, I have rigorously mastered Artificial Intelligence, Generative AI, Machine Learning, Data Science, Deep Learning and MLOps through countless hours of hands-on practice, online courses, and real-world projects. Driven by relentless curiosity and ambition, I’ve built production-ready, end-to-end systems across domains like Machine Learning, Deep Learning, Natural Language Processing, Data Science, Generative AI, and Cloud-based MLOps. My portfolio reflects practical expertise, industry-level implementation, and a passion for solving complex, high-impact problems with technology.
My Skills
A collection of the technologies and tools I'm proficient in, from programming languages to MLOps frameworks.
- -Scikit-learn
- -XGBoost
- -Data Science (Statistical Analysis, Predictive Modeling, Data Visualization)
- -Exploratory Data Analysis (EDA)
- -TensorFlow
- -Keras
- -PyTorch
- -Transfer Learning
- -Pandas
- -NumPy
- -Apache Airflow
- -Apache Spark
- -SQL (basic)
- -Docker
- -Kubernetes
- -MLflow
- -ZenML
- -GitHub Actions
- -CI/CD workflows, model versioning, monitoring
- -Amazon Web Services (AWS)
- -Google Cloud Platform (GCP)
- -Firebase (Auth, Firestore, Storage)
- -Python
- -SQL (basic)
- -Bash
- -Matplotlib
- -Seaborn
- -Streamlit (interactive dashboards)
- -Gradio (AI model demos)
- -GPT
- -BERT
- -BART
- -Tokenization
- -Attention Mechanisms
- -Named Entity Recognition (NER)
- -PEFT
- -LoRA
- -RAG
- -LangChain
- -ChromaDB
- -AI agents
- -Hugging Face Transformers
- -FastAPI
- -Flask
- -HTML/CSS/JS (basic)
- -RESTful APIs
Projects
Showcasing my best production-ready AI systems, Machine Learning, Deep Learning, and NLP projects.
Production-Ready AI Systems
AI-powered resume platform to analyze, optimize, and generate ATS-friendly PDF resumes. Built with Next.js, FastAPI, Firebase (Auth, Firestore, Storage), and Google Cloud Run. Features include instant PDF feedback, job match scoring, secure authentication, and seamless CI/CD with GitHub Actions. Scalable, production-ready, and 100% free for users.
Impact: Empowers users to create and optimize ATS-friendly resumes, enhancing job application success through AI-driven formatting and keyword optimization.
End-to-end advanced MLOps pipeline for Telecom Churn Detection using ZenML, MLflow, and GitHub Actions – with automated data ingestion, preprocessing, training, model evaluation, drift detection, retraining, CI/CD, Dockerized deployment on AWS EC2, and DynamoDB integration.
Impact: Ensures continuous model performance with automated retraining and monitoring.
A scalable, end-to-end data pipeline for e-commerce review sentiment analysis using Apache Airflow, Spark, Firebase Firestore, and Streamlit. Automates ETL, sentiment scoring, and dashboard visualization in a fully containerized, production-ready workflow.
Impact: Delivers actionable insights for e-commerce platforms with real-time dashboards.
Machine Learning Projects
Predicts destination countries of new Airbnb users using their initial sign-up data. Implements data preprocessing, feature engineering, and classification with Logistic Regression and XGBoost. Includes performance evaluation and visual insights.
Impact: Achieves accurate predictions for user destinations, aiding personalized recommendations.
Predicted item sales at Big Mart using XGBoost regression. Handled missing values, encoded categories, and evaluated model performance with R² scores. The project applies supervised learning to forecast sales based on product and outlet features.
Impact: Forecasts sales with high R² accuracy for inventory optimization.
Built a sentiment analysis model using Multinomial Naive Bayes on the IMDB dataset. Applied TF-IDF vectorization, text cleaning, and hyperparameter tuning with GridSearchCV. Evaluated performance using accuracy, confusion matrix, and classification metrics.
Impact: Achieves high accuracy in classifying movie reviews as positive or negative, demonstrating effective text classification techniques for sentiment analysis applications.
Deep Learning Projects
A neural network-based credit risk scoring system using PyTorch to classify loan statuses as 'Fully Paid' or 'Charged Off.' Includes full data preprocessing, feature scaling, encoding, training pipeline, and performance evaluation using key classification metrics.
Impact: Classifies credit risk effectively, reducing potential losses.
A deep learning project using ResNet-50 in PyTorch to classify 120 dog breeds from the Stanford Dogs Dataset. Includes training pipeline, evaluation, and a Flask web app for real-time image-based predictions.
Impact: Achieves high accuracy in breed classification for real-time applications.
A comprehensive face recognition system with face enrollment, real-time webcam recognition, and Flask web interface. Uses OpenCV and face_recognition library with local JSON storage for face data. Supports both CLI and browser-based interaction with real-time webcam recognition capabilities.
Impact: Provides accurate face recognition with real-time webcam capabilities and web deployment, demonstrating effective computer vision techniques for biometric identification applications.
Natural Language Processing (NLP) Projects
A multi-source RAG chatbot using Mistral-7B, LangChain, and Gradio to answer queries from PDFs, YouTube, web, and Wikipedia. Includes console and web UI, supports memory, and is ideal for research, education, and general knowledge applications.
Impact: Provides accurate, context-grounded responses from multiple sources.
Modular AI-powered data analysis tool using a multi-agent architecture with Mistral-7B. Automates data cleaning, EDA, visualization, and report generation via a Gradio UI. Supports natural language queries for seamless, intelligent data exploration.
Impact: Automates full data analysis pipelines, saving hours of manual work.
Emotion detection using BERT transformer model, classifying text into six emotions with 92% accuracy. Features weighted loss, TensorBoard tracking, and Flask web deployment. Trained on Hugging Face Emotion dataset with advanced fine-tuning and evaluation.
Impact: 92% accuracy in emotion classification for sentiment analysis applications.
Achievements & Certifications
A testament to my commitment to continuous learning and professional growth in the field of AI.
Completed under Prof. Andrew Ng, covering supervised learning, regression, classification, and ML engineering practices.
View CertificateMastered neural networks, CNNs, RNNs, LSTMs, Transformers and sequence models, taught by Andrew Ng.
View CertificateGained hands-on experience in building and deploying generative AI applications using foundational models, prompt engineering, and enterprise AI tools.
View CertificateBuilt over 30 production-ready AI projects and scalable systems.
Continuously upskilling through platforms like Coursera, Udemy, and staying current with AI research papers.
Blog & Insights
Sharing knowledge on AI concepts, project lessons, and trends.
The AI world moves fast. Every week, there's a new framework promising to make your life easier. "Just drag and drop, and voilà, you've built an agent!" Tempting? Sure. But here's the thing: when you jump straight to these higher-level tools, you miss the most important part—understanding how AI agents work under the hood. That understanding is what sets apart those who use AI from those who can truly build it.
I have to share something that amazed me this week. Google DeepMind just introduced Genie 3, and I can't stop thinking about it. It's one of those moments when you read about a new technology and feel like you're getting a sneak peek into the future. There's awe and excitement, along with a nagging worry.