Devansh Kaushik

Built enterprise AI claim analyzer targeting €200k annual savings and 50% processing-time reduction. Designed GenAI workflows with LangChain, Azure OpenAI, RAG, and agent architectures.

Enhanced data-driven workflow platform supporting 500+ production planners with PEM Clarity agentflow. Developed and optimized REST APIs reducing latency by 60%.

Building enterprise AI applications and production software systems with focus on Generative AI and backend engineering.

Developed and evaluated binary fabric-stain classification models using pretrained SqueezeNet and GoogLeNet, achieving 89.83% test accuracy. Built a Python-based CPU benchmarking harness to measure inference latency and throughput.

Developed AI data pipelines converting raw enterprise data into standardized Silver data products. Built multi-agent Azure OpenAI pipelines for data ingestion and schema understanding.

Deployed ML pipelines with multi-model inference for invoice content segmentation. Implemented address detection with ensemble NLP model and Flask REST APIs.

Developed Stain Image Classification with SqueezeNet & GoogleNet achieving 96% training accuracy with a 26 MB model. Compared object detection models for fridge object segmentation.

Grade: 7.81

Pursued a Master of Technology in Computer Science and Engineering, achieving a CGPA of 7.81.

Grade: 8.97

Completed a Bachelor of Technology with Honours in Computer Science and Engineering, achieving a CGPA of 8.97.

Grade: 8.4

Earned a Bachelor of Science in Programming and Data Science with a CGPA of 8.4.

Grade: 93.2%

Completed Senior Secondary education with a percentage of 93.2%.

Grade: 9.8

Completed Secondary education with a percentage of 9.8.