Riaz Ahmed. T

Github

Contact

About myself

• Dynamic and forward-thinking DevOps/AWS Cloud Architect and Big Data Platform leader with a proven track record of over 22 years in the global IT industry, spanning USA, Singapore, and India.
• Hands on experience in designing, architecting, and managing large-scale, complex AWS cloud based and big data solutions with a strong emphasis on security and architectural integrity.
• Currently expanding focus toward Datascience, Generative AI and Machine Learning, with a strong interest in applying cloud-native tools and MLOps best practices to deliver scalable, intelligent solutions.
• Successfully recruited, trained, and led a team of 50 full-time employees across NAM and APAC regions, establishing a robust global support model while fostering team cohesion and effectively resolving conflicts.
• Demonstrated ability to excel both as a collaborative team player and a self-reliant individual contributor. Possess exceptional client-facing skills and a keen aptitude for independently mastering new technologies, delivering innovative solutions that generate significant business value for clients.

Industry Certifications

• Amazon Web Services (AWS) Certified Solution Architect
• Certified Kubernetes Application Developer (CKAD)
• HashiCorp Certified Terraform Associate
• Databricks Certified Data Engineer Associate
• Databricks Certified Associate Developer for Apache Spark 3.0
• Associate Data Analyst with Python Track – DataCamp
• Oracle Cloud Infrastructure 2019 Certified Architect Associate
• Cloudera Certified Administrator for Apache Hadoop (CCAH)
• Digital Transformation and Design Thinking for Innovation – University of Virginia & Boston Consulting Group (Coursera)
• Microsoft Certified System Engineer Windows 2000 (MCSE) with Messaging Specialization

Datascience projects

1. House Price Prediction using XGBoost and ANN.

• Real estate value prediction model.
• Data source: Kaggle - Housing Prices.
• Includes feature engineering, visualization, XGBoost & ANN modeling.
• GridSearchCV & RandomizedSearchCV for hyperparameter tuning.
• Final model recommendation based on metrics comparison.

2. Predicting Neurodegenerative Disease using KNN and PCA.

• Prediction based on test results like Albumin, Sodium, Urine PH, etc.
• Data source: Bellevue University.
• Used clustering, silhouette analysis, PCA for visualization.
• Eigenvalues and eigenvectors calculated post PCA.

3. Sentiment Analyzer using NLTK.

• Movie review sentiment prediction using TextBlob and Vader.
• Includes tokenization, stopword removal, stemming.
• TFIDF Vectorization and models: Logistic Regression, SVM.
• Final model chosen based on evaluation metrics.

4. Breast cancer prediction using linear model in R language.

• Predicting the chances of breast cancer depending upon the age, tumour size, characteristic, menopause stage etc.
• Data Source: Oncology institute.
• Used R language, ggplot, tidyverse, broom, caret and metrics package.
• Converting the variables to factor.
• Used glm - linear model and stepAIC function to determine the optimum predictor variables.

5. Predicting customer churn for subscription-based program.

• Predicting whether the customer will churn or not, by using the existing data of the subscription provider.
• Data Source: Kaggle, https://www.kaggle.com/datasets/safrin03/predictive-analytics-for-customer-churn-dataset.
• Used Chi Squared test of independence to compare how the churn is associated with other variables.
• Used Standard scaler, One-Hot encoding and SMOTE method has been used.
• Dimensionality reduction using PCA.
• Analysis of linear separation of data based on Principal components and comparison of PCA vs RBF.
• Model Building using RandomizedSearchCV.

6. Implementing Retrieval Augmented Generation (RAG) for LLM model.

• RAG implementation using ChromaDB for gpt-4o-mini openai model.
• Used langchain and Wikipedia libraries.
• RecursiveCharacterTextSplitter has been used for splitting.
• OpenAIEmbeddings class has been used to generate vector embeddings.

7. Personalized gift recommendation GenAI model and streamlit.

• Fine tuning a personal GenAI model using gpt-4o-mini to offer personalized gift suggestions.
• OpenAI model has been used and fine tuning job has been created using an uploaded training file.
• Fine tuning job information retrieval and metrics evaluation.
• Model inference and deployment using Streamlit.

8. Pulsar star prediction using ANN and RandomForest models.

• Supervised binary classification algorithm to determine if the reading of star is for Pulsar or not.
• Data Source: https://archive.ics.uci.edu/dataset/372/htru2
• Artificial Neural Network model using Keras.
• Hyperparameter tuning with keras tuner.
• Model comparison and recommendation between ANN and Random Forest Classifier model.

9. Effect of music on emotions using ANN and KNN models.

• Supervised multiclassification algorithm, predicting the emotions based on different parameters of Turkish music.
• Data Source: https://archive.ics.uci.edu/dataset/862/turkish+music+emotion
• ANN model has been built with Pytorch. Optuna libraries has been used to do hyperparameter tuning.
• Additional KNN model has been built, Cross Validation method used to find optimum K value
• Model comparison based on metrics and final recommendatation

10. Flight Price Prediction using ANN and Random Regressor models.

• Supervised regression algorithm to predict the flight price based on various input features like Airline, Source, Destination etc.
• Support Vector Machine with Radial Basis Function, ANN with Keras library and Random Forest Regressor models has been built.
• Effect of Anova F test and PCA on the model
• Model comparison based on metrics and final recommendatation.