Shiva Thavani - AI Engineer

machine learning engineer · ask me anything

GenAI • Agentic AI • LLMOps • MLOps · Building production-grade AI systems, enterprise ML platforms, and scalable LLM-powered applications.

try:
15+ Production Systems Shipped
5+ Years end-to-end ML Exp
$500K+ GPU/Infra Costs Optimized
40% Complex Support Resolution
scroll for projects ↓

About

Enterprise AI, engineered for production.

Shiva Thavani - Senior AI & Machine Learning Engineer

Shiva Thavani

GenAI • Agentic AI • LLMOps • MLOps

AI/ML Engineer with 5+ years of end-to-end ML experience building enterprise-scale AI systems, LLM applications, and MLOps platforms across Atlassian, Autodesk, Thomson Reuters, and EY.

Specialized in Agentic AI workflows, GenAI deployment, LLM serving, GPU optimization, and scalable ML platforms across 3 major cloud providers.

15+

Production Systems Shipped

5+

Years end-to-end ML Exp

$500K+

GPU/Infra Costs Optimized

40%

Complex Support Resolution

Experience

AI platforms across high-scale product teams.

Atlassian

Machine Learning Engineer

Jun 2025 - Present
  • Achieved ~40% reduction in mean time to resolution (MTTR) via CSM-Rovo agentic incident workflows
  • Automated 4 core support and engineering workflows via CSM-Rovo integration
  • Integrated 2 frontier LLMs (GPT-5.2 + Gemini-3-Flash) utilizing 1M+ token context windows via GraphRAG pipelines
  • Deployed 5+ GenAI models across standard product suite
LangGraph MCP GraphRAG LlamaIndex OpenAI Gemini

Autodesk

Senior AI/ML Engineer

Jul 2024 - Jun 2025
  • Optimized GPU and infrastructure costs by ~$500K+ (representing 30% of enterprise GPU spend)
  • Served 4 enterprise applications via robust AI Inference Service on AKS & Docker
  • Orchestrated MLOps tools (Terraform + MLflow + Kubeflow) across Azure
  • Achieved ~40% reduction in infrastructure deployment time via Terraform IaC automation
AzureML CUDA Kubeflow MLflow Terraform Docker AKS

Thomson Reuters

Senior AI Engineer

Sept 2022 - Jul 2024
  • Built enterprise AI platform supporting 200+ ML engineers across AWS & Azure, integrating 8+ cloud services
  • De-risked and deployed 4 LLMs in production (LLaMA2, LLaMA3, Claude3, GPT-4)
  • Orchestrated 8+ AWS/Azure cloud services to streamline production-grade LLM deployments across the organization
  • Managed 2-year platform ownership from early development to full organization adoption
AWS Azure SageMaker Kubernetes Datadog

EY GDS

Data Scientist

Apr 2021 - Aug 2022
  • Developed OCR and Document AI systems deployed in cross-platform production environments (ONNX + TFLite)
  • Achieved ~10x faster document processing vs manual review with 90%+ extraction accuracy
  • FTE conversion within 1 year from intern, validating strong quality delivery
TensorFlow OpenCV ONNX TFLite Document AI Python

Skills

Full-stack AI delivery, from prompts to GPU clusters.

Generative AI

LangGraph LangChain MCP GraphRAG LlamaIndex OpenAI Claude Gemini RAG VectorDB

AI/ML

PyTorch TensorFlow CUDA ONNX Hugging Face OpenCV

MLOps

Kubeflow MLflow SageMaker AzureML Docker Kubernetes Terraform

Cloud

AWS Azure GCP

Backend

Python FastAPI Flask Golang SQL

projects.

6 production systems and highlighted builds

Agentic AI Customer Support System

Enterprise AI orchestration system using LangGraph + MCP with GraphRAG retrieval.

45% reduction in manual intervention
10k+ AI-assisted conversations
LangGraph MCP GraphRAG

Multi-Cloud Enterprise MLOps Platform

Complete MLOps lifecycle platform covering end-to-end model development, scalable pipeline deployment, and production inference/monitoring supporting 200+ ML engineers across multiple clouds.

~$500K+ GPU costs optimized
2.5x throughput improvement
200+ ML engineers supported
AWS Azure GCP SageMaker Kubeflow Kubernetes Terraform MLflow CUDA
Smart AI Robot - Shiva Thavani AI Project

Smart AI Robot

AI robotics hackathon project recognized at Bosch AI Hackathon 2021 for smart robot automation.

1st Prize Winner
Theme: Smart AI Robot
AI Robotics Automation
Open Innovation Prototype - Shiva Thavani AI Project

Open Innovation Prototype

National innovation project recognized at Smart India Hackathon 2020.

Winner
Theme: Open Innovation
AI Product Engineering Rapid Prototyping

Achievements

Hackathon wins and academic distinction.

Winner Smart India Hackathon 2020

Winner Smart India Hackathon 2020

Theme: Open Innovation

✦

EY GDS Hackathon Winner

Theme: Intelligent Automation with AI

Bosch AI Hackathon Winner

Bosch AI Hackathon Winner

Theme: Smart AI Robot

✦

Xebia Xe-Thon Winner

Theme: Intelligent hiring using AI

Gold Medalist

Gold Medalist

Academic BE-MBA degree qualification

✦

Merit Scholarship Holder

TIET merit scholarship holder for 2018-2019 and 2019-2020

Publications

Peer-reviewed research.

Published research in computer vision and AI.

📄

Pose Invariant Non-Frontal 2D, 2.5D Face Detection and Recognition Technique

International Journal of Information Technology

Springer · 2023 4 Citations

Shiva Thavani, Sahil Sharma, Vijay Kumar

Proposed the PIIR technique using facial landmarks and 3D Morphable Models for pose-invariant face recognition. Achieved 90.24% authentication accuracy on CAS-PEAL dataset at 30° pan-angle.

Computer Vision Face Recognition 3D Morphable Models Deep Learning
Read on Springer →

DOI: 10.1007/s41870-023-01335-2

Contact

Let’s build production AI.

Reach out for GenAI platforms, agentic workflows, LLMOps systems, or scalable inference engineering.

Email thavani.shiva3@gmail.com LinkedIn shiva-thavani GitHub shiva2410
Download Resume Start a Conversation