Devi Kathirvel

Software Engineer | AI & ML Developer

Chennai, India 1+ yrs exp 83 · Excellent

About

Software Engineer specialising in AI/ML with 1.5+ years of experience building production-grade document intelligence systems. Experienced in GPU-accelerated OCR pipelines, open-source LLM deployment, and LayoutLMv3-based document understanding. Delivered measurable results across extraction accuracy, processing throughput, and API response times. Comfortable across the full stack from model training to cloud deployment.

Skills & Expertise (24)

Python Advanced

7.8/10

1.5

Years Exp

FastAPI Advanced

7.5/10

1.5

Years Exp

Deep Learning Advanced

7.0/10

1.5

Years Exp

Git Advanced

7.0/10

1.5

Years Exp

Docker Advanced

7.0/10

1.5

Years Exp

Anomaly Detection Advanced

7.0/10

1.5

Years Exp

NLP Advanced

7.0/10

1.5

Years Exp

scikit-learn Advanced

7.0/10

1.5

Years Exp

TensorFlow Advanced

7.0/10

1.5

Years Exp

RESTful API Design Advanced

7.0/10

1.5

Years Exp

Flask Advanced

7.0/10

1.5

Years Exp

LayoutLMv3 Intermediate

6.5/10

1.5

Years Exp

Azure DevOps Intermediate

6.5/10

1.5

Years Exp

LINUX Intermediate

6.5/10

1.5

Years Exp

AWS Intermediate

6.5/10

1.5

Years Exp

Azure Intermediate

6.5/10

1.5

Years Exp

MongoDb Intermediate

6.0/10

1.5

Years Exp

Heroku Intermediate

6.0/10

1.5

Years Exp

MySql Intermediate

6.0/10

1.5

Years Exp

Postgresql Intermediate

6.0/10

1.5

Years Exp

Ollama Intermediate

6.0/10

1.5

Years Exp

JavaScript Intermediate

5.5/10

1.5

Years Exp

CSS Intermediate

5.0/10

1.5

Years Exp

HTML Intermediate

5.0/10

1.5

Years Exp

Work Experience

Software Engineer – Python | AI | ML

Digital Hoop Technologies Pvt. Ltd.

Dec 2025 - Apr 2026

Built a GPU-accelerated OCR pipeline on RunPods, cutting document inference time by ~45% and processing 8,000+ pages per day. Deployed open-source LLMs via Ollama for on-premise content understanding, eliminating external API costs and reducing response latency by 30%. Fine-tuned LayoutLMv3 model for document layout analysis, achieving 87% extraction accuracy on structured PDFs and scanned forms — up from a 61% baseline. Containerized 5+ AI microservices with Docker, reducing environment setup time from hours to under 10 minutes. Built FastAPI endpoints to serve AI models in production, supporting 200+ concurrent requests with <300 ms average response time. Managed end-to-end deployment to IRS remote servers via Azure DevOps CI/CD, reducing manual release effort by 60%.

Software Developer – Python | AI | ML

EzeeShipping Solutions

Sep 2024 - Nov 2025

Built anomaly detection models that flagged irregular shipment patterns with 91% precision, reducing manual audit effort by 35%. Designed Flask RESTful APIs handling 500+ daily requests for shipment tracking and data exchange across 3 internal systems. Developed an NLP-based intent classification chatbot that resolved 70% of user queries without human escalation, handling 300+ sessions/day. Applied feature engineering and hyperparameter tuning to predictive models, improving forecast accuracy by 18%. Deployed ML models on AWS and Azure, maintaining 99.5% uptime across production workloads. Reduced data preprocessing pipeline runtime by 40% through vectorised Pandas operations and optimised SQL queries.