About Me

Shivani Gowda KS

Hey there, I’m the kind of engineer who enjoys understanding why things work just as much as how they work. My career has taken me through AI/ML and Azure networking (L3/L7), where I learned to make systems both intelligent and reliable.

These days, I’m knee-deep in personal projects around ONNX, C++ inference, and model optimization, chasing performance improvements the way some people chase coffee refills. I’m not a fast learner by nature; I prefer to slow down, get the fundamentals rock solid, and then move fast with confidence. Once something clicks, I dive all in. I love the blend of research and engineering, building ML tools, experimenting with agents, and learning how to make complex ideas work in the real world. My favorite feeling is when something I built finally runs just right and I immediately start thinking about how to make it better. If you’re into thoughtful engineering, LLM experiments, or the art of making models a little smarter (and a lot faster), we’ll probably get along.

Professional Experience

Cloud Engineer (Vendor to Microsoft)

LTIMindtree · Bellevue, WA, USA
March 2024 – Present

Details to be added.

Machine Learning Engineer (Contract)

Flavor · Remote, USA
Nov 2023 – Jan 2024

Tools: Lorem ipsum dolor sit amet consectetur adipisicing elit. Molestias vel voluptatibus sit odit architecto tempore praesentium dignissimos possimus aperiam ducimus dolore neque, unde voluptatum obcaecati iusto! Suscipit quidem unde eligendi?

Lorem ipsum dolor sit amet consectetur adipisicing elit. Inventore ducimus, neque facilis animi iusto corrupti aliquam doloremque! Optio quisquam pariatur sunt aliquid vel, dolore beatae, excepturi, iure soluta in quidem.

Machine Learning Engineer (Intern → Full-time)

PixStory · Remote, USA
Jan 2023 – Dec 2023

Tools: Python, Flask, SQLite, Apache Airflow, BERT, NLLB, Transformers

Built ETL pipelines with Airflow (31% performance boost) and a Flask/SQLite annotation platform with RESTful APIs for real-time labeling (15% increase in data collection). Fine-tuned multilingual transformer models for hate speech classification and benchmarked embedding models for unstructured data indexing.

Graduate Research Assistant

Loyola Marymount University · Los Angeles, CA, USA
Jun 2022 – Dec 2022

Tools: PyTorch, Transformers, Data Processing

Improved diet tracking system performance by 21.8% by replacing RNN+LSTM with transformer architecture. Collected and analyzed large datasets, identified bottlenecks, and boosted F1-score by 5.2%. Published a paper at ICASSP 2023 on multi-modal food classification.

Application Developer

IBM · India
Dec 2019 – May 2021

Tools: Java, Spring Boot, OpenShift, GitLab, Jenkins

Developed backend logic for a CPQ system using Java and Spring Boot. Assisted in migrating legacy services to cloud-native microservices on OpenShift and used CI/CD pipelines (GitLab, Jenkins) to execute deployments and run validation scripts.

Education

Loyola Marymount University

Master of Science in Computer Science
Aug 2021 – May 2023

Los Angeles, CA, USA. Received 'Outstanding Graduate Award in Computer Science' from Frank R. Seaver College of Science and Engineering.

Amrita School of Engineering

Bachelor of Technology in Electrical and Electronics Engineering
Jun 2015 – May 2019

Bengaluru, KA, India.

Skills & Technologies

Programming Languages

Python C C++ Java Bash SQL

AI / ML

PyTorch Hugging Face ONNX LangChain Pandas scikit-learn NumPy

Tools & Platforms

Linux Git Wireshark Postman PySpark Grafana

Databases

SQLite MySQL

Cloud & DevOps

Azure AWS OpenShift Docker Jenkins

Web Development

Flask FastAPI Spring Boot HTML5 CSS3 JavaScript React

Projects

Some of the projects I've worked on in cloud engineering and automation

Multi-Modal Food Classification in a Diet Tracking System

PyTorch Vision Speech

Leveraged multi-modal deep learning techniques (vision + spoken inputs) to classify food items for a diet-tracking system. Built data pipelines and preprocessing for large-scale training.

Nichirin: Webcrawler + Retrieval-Augmented Generator

Apache Solr Apache Spark Python

Multi-level web crawler with Solr-powered indexing and retrieval; integrates Spark for parallel processing and scalable text indexing. Open sourced and published to PyPI.

Azure Network Automation

View on GitHub Live Demo
Azure Python PowerShell ARM Templates Logic Apps

Developed a comprehensive automated network provisioning system that revolutionizes how enterprise Azure deployments are managed. This solution reduces manual configuration time by 80% and significantly improves reliability through automated monitoring and alerting systems.

Key Features & Achievements:

  • Automated VNet and subnet provisioning with intelligent IP allocation
  • ExpressRoute configuration automation with failover capabilities
  • Real-time network monitoring dashboards with custom metrics
  • Automated incident response workflows reducing MTTR by 60%
  • Cost optimization recommendations saving 30% on network resources
  • Integration with existing ITSM tools for seamless operations

Multi-Modal Food Classification in a Diet Tracking System

GitHub Repo Project Details
PyTorch TensorFlow Vision Speech Data Pipelines

Leveraged multi-modal deep learning (vision + spoken inputs) to accurately classify food items for a diet-tracking system. As part of this work I gathered large-scale datasets, engineered preprocessing pipelines, and trained transformer-based models to improve classification accuracy.

Key Achievements:

  • Applied multi-modal architectures for image+speech inputs to improve food classification accuracy.
  • Built scalable data collection and preprocessing pipelines; created a web scraper to gather training data at scale.
  • Improved model robustness and F1 scores through careful dataset cleaning and augmentation.

Nichirin: Webcrawler + Retrieval Augmented Generator

GitHub Repo PyPI (pip install nichirin)
Apache Solr Apache Spark Python Scrapy Selenium

Nichirin is a multi-level web crawler and retrieval-augmented generator that uses Apache Solr for text indexing and retrieval. It integrates Apache Spark for parallel processing to scale web crawling and indexing, and includes preprocessing steps to structure raw inputs for downstream embedding and retrieval.

Platform Features:

  • Multi-level crawling with configurable depth and rate limits.
  • Apache Solr-based indexing for fast full-text retrieval.
  • Spark integration for parallel processing and scalable indexing.
  • Preprocessing strategies to clean and structure raw HTML/text for embeddings.
  • Published to PyPI (pip install nichirin) and open sourced on GitHub.

Security Compliance Automation

View on GitHub Security Report
Azure Policy Security Center PowerShell Log Analytics Sentinel

Developed a comprehensive automated system for monitoring security compliance across enterprise Azure environments with intelligent remediation capabilities. This solution ensures continuous compliance and dramatically reduces security risks.

Security Impact:

  • 99% compliance score achievement across all enterprise environments
  • Automated remediation of 500+ policy violations daily
  • Real-time security posture monitoring and reporting
  • Reduced manual audit time by 90% through automation
  • Custom security policies tailored to industry requirements
  • Integration with SIEM tools for enhanced threat detection

DevOps Pipeline Optimizer

View on GitHub Documentation
Azure DevOps CI/CD YAML PowerShell Docker

Built an intelligent pipeline optimization tool that analyzes build and deployment patterns to automatically suggest and implement efficiency improvements, reducing overall CI/CD cycle times significantly.

Key Features:

  • Automated pipeline analysis and bottleneck detection
  • Smart caching strategies reducing build times by 50%
  • Parallel execution optimization across multiple agents
  • Automated test suite optimization and selection
  • Cost optimization through efficient resource utilization
  • Integration with Azure Boards for automated tracking

Cloud Cost Analytics Dashboard

View on GitHub Live Demo
Power BI Azure Cost Management Python Azure Functions SQL

Developed a comprehensive cost analytics platform providing real-time insights into cloud spending patterns, forecasting future costs, and identifying optimization opportunities across multi-cloud environments.

Dashboard Capabilities:

  • Real-time cost tracking across Azure, AWS, and GCP
  • AI-powered cost forecasting and trend analysis
  • Automated anomaly detection for unexpected spending spikes
  • Department and project-level cost allocation
  • Actionable optimization recommendations saving 25% on average
  • Customizable alerts and automated reporting

Latest Posts

Thoughts on technology, cloud engineering, and life experiences

View All Posts

Contact

Let's connect - reach out via email, LinkedIn, or check out my work on GitHub.

LinkedIn GitHub

Beyond the Code

Want to know more about me? Welcome to my personal space, a little window into the world outside the terminal.

Explore my personal space