Ranjana Rajendran

Applied Machine Learning and Software Engineer

Production ML Systems | LLM Fine-tuning & Deployment | MLOps Infrastructure | Generative AI | Big Data

Seattle, WA, USA | Authorized to work in US without sponsorship

About Me

Software Engineer with 10+ years of experience building production ML infrastructure at companies including AWS and Mastercard. MS in Computer Science from UC Santa Cruz with deep foundations in Machine Learning, Data Mining, and Mathematical Statistics. Strong track record in MLOps, release engineering, and supporting data scientists deploying models at scale.

Passionate about the mathematical foundations of machine learning and bringing rigorous engineering practices to AI systems. Currently focused on LLM deployment challenges: context management, retrieval systems, and infrastructure for serving models at scale.

Seeking full-time ML/MLOps engineering roles where I can apply software engineering rigor and mathematical reasoning to production AI challenges.

Patent Holder

US-20250053985-A1: Cryptocurrency Fraud Detection

Published Researcher

Data Mining and Knowledge Discovery, Volume 28

Certifications

Cloudera CCDH, CCAH, CCSHB

Work Experience

Independent ML Engineer & Author

August 2023 - Present
  • Authored "Engineering Autonomous AI" - comprehensive guide to building production agentic systems
  • Built and deployed Tech Stack Advisor - production multi-agent AI system with 5 specialized agents, LangGraph orchestration, and full CI/CD pipeline
  • Mastered LLM deployment, multi-agent systems, and production AI infrastructure through hands-on projects
  • Implemented LLMOps infrastructure including Prometheus/Grafana monitoring, JWT authentication, and Railway deployment
  • Developed deep learning projects: VAE for generative modeling, BERT/Longformer for NLP, traditional ML for regression and clustering

Lead Software Engineer - Ekata-Mastercard

August 2022 - August 2023
  • Engineered distributed ETL pipelines over terabytes of identity data using Spark on Databricks & EMR
  • Managed Redis-backed identity graph supporting real-time ML fraud scoring
  • Co-invented patented fraud detection system (US-20250053985-A1)

System Development Engineer - Amazon Web Services

July 2017 - August 2022
  • Release engineer for Elastic Map Reduce (EMR), maintaining 99.9% service uptime across multiple releases
  • Implemented and maintained integration testing framework (200+ automated tests), reducing release-blocking bugs by 60%
  • Resolved 100+ critical integration issues between ML and big data frameworks (Hive, Spark, Presto), reducing customer-impacting incidents by 40%
  • Supported EMR deployments processing petabytes of data for thousands of enterprise customers

Senior Developer Support Engineer - Qubole

April 2016 - May 2017

Optimized large-scale data and feature engineering pipelines, achieving up to 70% faster runtimes and 90% fewer pipeline failures.

Hadoop Engineer - Altiscale

February 2015 - March 2016
  • Handled customer feature requests and support tickets for production Map-Reduce/Spark jobs, ML data pipelines, ETL, and security (Kerberos)
  • Troubleshot, optimized, and tuned customer Hive, Spark, and Spark SQL applications
  • Java performance and memory profiling using Eclipse memory analyzer tool
  • Converted customer ML pipeline needs to feature requests and authored knowledge-base articles

Solution Architect - Cloudera

September 2013 - January 2015

Deployed and managed Hadoop ecosystem supporting large-scale ML training pipelines and inference workloads.

Machine Learning Projects

Explore my hands-on ML projects covering traditional ML, deep learning, and generative AI

Generative AI

Tech Stack Advisor (Live Production)

Production-ready multi-agent AI system for intelligent tech stack recommendations with conversational agent that actively seeks clarifications through multi-turn dialogue. Features short and long-term memory using Qdrant vector store for context-aware conversations. End-to-end implementation with CI/CD pipeline, Prometheus/Grafana monitoring, LangGraph orchestration, FastAPI REST API, JWT authentication, and Railway deployment. 5 specialized AI agents working in concert.

Production Multi-Agent AI CI/CD Prometheus Grafana LangGraph Qdrant FastAPI Docker
View Project

MLOps & Infrastructure

ML Model Serving Protocol Comparison

Comprehensive performance benchmark comparing REST, gRPC, and GraphQL for serving ML models in production. Built complete testing infrastructure with Locust load testing, Prometheus/Grafana monitoring, and automated result visualization. gRPC demonstrated 30x lower latency and 3x higher throughput. Features interactive architecture diagrams and detailed performance analysis across 8 metrics.

MLOps Performance gRPC REST GraphQL Docker Prometheus Locust FastAPI
View Project

Deep Learning

Group Emotion Recognition

Computer vision system for analyzing collective emotions in group photographs using face detection (RetinaFace), emotion recognition (DeepFace), and quality-weighted aggregation. Built on Google Vertex AI. Features entropy-based emotional diversity analysis.

Computer Vision Vertex AI RetinaFace DeepFace
View Project

Spoiler Detection (NLP)

Deep learning system for automatic spoiler detection in movie reviews using transformer models (BERT, Longformer) and LSTM. Achieved 71.5% accuracy with Longformer's extended context window handling 4096 tokens.

NLP BERT Longformer PyTorch
View Project

Traditional ML

Volcano Eruption Prediction

Time series regression for volcanic activity forecasting using sensor data. Achieved 99.7% R² with RidgeCV for pressure prediction and 95.9% R² with Random Forest for time-to-eruption. Features tsfresh feature extraction and sliding window analysis.

Time Series Regression tsfresh scikit-learn
View Project

Social Network Clustering

Community detection and clustering on Facebook social network (4,039 nodes, 88K edges). Achieved 0.73 silhouette score with Agglomerative Clustering. Features graph embeddings with Node2Vec, 20+ graph-based features, and comprehensive algorithm comparison.

Graph ML Clustering Node2Vec NetworkX
View Project

Technical Skills

Cloud & DevOps

AWS Railway Docker Kubernetes Jenkins GitHub Actions Git Prometheus Grafana SSL/HTTPS

Machine Learning

PyTorch TensorFlow Scikit-Learn HuggingFace Langchain LangGraph Anthropic Claude Pydantic RAG Transformers LLM

Big Data

Spark Hadoop Hive Presto Airflow EMR Databricks

Programming

Python Java C/C++ Scala R SQL

Data Visualization

Streamlit Plotly Matplotlib Dash

Databases

MySQL SQLite Redis Qdrant FAISS HBase Cassandra

Web Development

FastAPI HTML5 CSS3 JavaScript REST APIs Swagger/OpenAPI

Testing & Quality

pytest mypy ruff structlog

Security & Auth

JWT OAuth 2.0 bcrypt Rate Limiting

GitHub Activity

Interactive 3D visualization of my recent commit history

Loading GitHub activity...

-

Recent Commits

-

Active Public Repositories

-

Active Days

Get In Touch

I'm open to discussing ML engineering opportunities, collaborations, and interesting projects.