building the AI Employees.

• Leading the ML team to develop deep learning based Invoice automation engine.• Research and development of custom LLM based applications with langchain.• Experience with LLM include, Fine-tuning with LoRa/QLoRa, Instruction tuning, Optimization with Llama.cpp,GPTQ,AWQ and efficiently deploying the LLM at scale.• Reduced the Inference time by 30 times in CPU and memory by 3 times by performing state-of-the art optimization techniques for the Invoice automation engine.• Led the successful deployment of the ML engine on AWS Fargate, orchestrated through CICD pipelines, enhancing system reliability and automating deployment processes.• Implemented Asynchronous processing and task orchestration using Amazon ECS, enabling seamless scalability and efficient utilization of resources.• Development of Semi automatic annotation tool to increase the speed of annotation by 20 times.• Development of recommendation model using one-shot learning and few-shot learning.

• Developed Cyber Risk scorecard model using Machine Learning• Involved in design and development of data pipelines in PySpark which includes, Data transformation, utlier removal, Data Appends and feature engineering• Performed statistical analysis to generate actionable insights• Implemented End to End ML model building workflow in Data bricks using MLFlow which includes, Feature selection, Hyperparameter tuning, Model Validation and Model selection• Actively collaborated and supported to US analytics team for various projects

- Developed various Deep Learning Computer Vision solutions for Industry safety and maintenance- Increased the inference speed of the model more than 10 times by using various optimizing techniques- Achieved more than 90% accuracy in the majority of the DL solutions- Majority of work include model building and optimizing using Tensorrt and Deploying on the edge devices like jetson platform- Have contributed to build the entire pipeline of the project

- Worked on Data cleaning and feature Engineering- Contributed to Model building of Machine Learning and NLP based projects- Contributed to build pypi package and docker images