Sai Chandu Machavarapu — MLOps Engineer

About

Bridging ML
& Production

I'm a Masters student in Computer Science at UT Tyler, graduating in 2026 — with a background that spans AI/ML engineering and DevOps infrastructure.

My work sits at the intersection of machine learning and production systems — building the pipelines, monitoring stacks, and deployment workflows that take models from experiment to reality. I've worked across AWS SageMaker, GCP, and Azure, and I care deeply about systems that are observable, automated, and built to last.

Outside of work, I'm driven by the belief that great ML infrastructure should be invisible — it just works, quietly, in the background while the science takes center stage.

Experience

Professional
Journey

Nov 2023 — Apr 2024

Codegnan

Internship

Machine Learning Intern

Built an end-to-end house price prediction model using Python, scikit-learn, and ensemble techniques with feature engineering and hyperparameter optimization.
Developed an interactive Streamlit web application for real-time price predictions with optimized preprocessing and API integration.
Automated a data pipeline using web scraping to collect, clean, and validate 5,000+ property records for model training.
Deployed the full project on GitHub with comprehensive documentation, improving prediction accuracy through ensemble modeling.

May 2023 — Oct 2023

Wipro

Internship

Java Full Stack Trainee

Built scalable backend services using Spring Boot, Hibernate/JPA, and MySQL with RESTful API design and Spring Security implementation.
Developed responsive frontend interfaces using HTML5, CSS3, JavaScript, and React/Angular with seamless backend API integration.
Created an HR management application featuring employee records, attendance tracking, leave management, and role-based access control.

Mar 2022 — May 2022

Amazon Web Services

Virtual Internship

AI/ML Intern

Developed and deployed ML models on AWS SageMaker for classification and regression tasks with optimized compute configurations.
Built automated ETL pipelines using AWS Lambda, S3, and Python for ingestion and preprocessing of structured and unstructured datasets.
Deployed scalable ML inference endpoints with CI/CD automation, CloudWatch monitoring, and logging for production-grade model serving.

Projects

Production
MLOps Systems

Credit Card Fraud Detection Pipeline

MLflow · FastAPI · Docker · AWS ECS · Prometheus · GitHub Actions

Production fraud detection with automated retraining, drift monitoring, and real-time inference on AWS.

End-to-end pipeline from raw transaction data to deployed fraud detection API serving real-time predictions.
MLflow experiment tracking with model registry and automated promotion — handles imbalanced data with SMOTE and class weighting.
Automated retraining via GitHub Actions when new labeled data arrives in S3, with accuracy threshold gate before promotion.
Prometheus metrics tracking prediction latency, fraud flag rate, and feature distributions with Grafana alerting.

GitHub

Customer Churn Prediction Platform

MLflow · FastAPI · Streamlit · Docker Compose · GCP Cloud Run

Self-service ML platform letting business teams run churn predictions without touching code — deployed on GCP.

Self-service churn prediction with Streamlit UI — upload CSV, get batch predictions, view model performance over time.
Containerized with Docker Compose (API + MLflow + monitoring) and deployed on Google Cloud Run.
Accuracy degradation alerts via Grafana when model performance drops below threshold.

GitHub

Sentiment Analysis API with Auto Retraining

DistilBERT · MLflow · FastAPI · PostgreSQL · GitHub Actions · AWS ECS

NLP inference API that self-improves by automatically retraining on its own low-confidence predictions.

NLP API for product review sentiment — logs low-confidence predictions to PostgreSQL as a feedback loop for retraining.
Weekly automated retraining via GitHub Actions using accumulated low-confidence samples.
MLflow Model Registry manages version promotion with automated staging → production gates.

GitHub

House Price Prediction with Feature Store

Feast · XGBoost · MLflow · FastAPI · Docker · GCP Cloud Run

ML system with a proper feature store architecture — eliminating training-serving skew from the ground up.

Feast feature store with offline (Parquet on S3) and online (Redis) stores — eliminates training-serving skew completely.
CI/CD pipeline runs automated tests, feature validation, and model benchmarks on every code push.
Feature distribution shift monitoring with Prometheus metrics and Grafana dashboards.

GitHub

Real-Time ML Monitoring Dashboard

Evidently AI · FastAPI · Prometheus · Grafana · Docker Compose · Slack

Full-stack ML observability with live drift detection, automated alerting, and one-command deployment.

Simulates live data streams with gradually drifting distributions to test production monitoring systems.
Evidently AI drift reports — KS-test triggers Slack webhook alert when drift exceeds threshold.
Grafana tracks model health, feature drift, latency percentiles, and prediction volume in real time.
Docker Compose — one command spins up the full API, monitoring, and alerting stack.

GitHub

Sai Chandu
Machavarapu

Bridging ML
& Production

Tools &
Technologies

Professional
Journey

Production
MLOps Systems

Academic
Foundation

Credentials &
Achievements

Let's
Connect

Sai ChanduMachavarapu

Bridging ML& Production

Tools &Technologies

ProfessionalJourney

ProductionMLOps Systems

AcademicFoundation

Credentials &Achievements

Let'sConnect

Sai Chandu
Machavarapu

Bridging ML
& Production

Tools &
Technologies

Professional
Journey

Production
MLOps Systems

Academic
Foundation

Credentials &
Achievements

Let's
Connect