Hi, I'm

Prakash Singh

About

I have prior experience working in deep natural language processing,Generative AI and computer vision, specifically with Document AI. I am fervently passionate about artificial intelligence. I possess a profound interest in comprehending the inner workings of AI algorithms. Consequently, I derive great satisfaction from implementing algorithms from scratch, as it allows me to gain a profound understanding of their intricacies. My ultimate goal is to tackle intricate problems within the realm of AI. I am currently specializing in reinforcement learning.

City: Chamoli, Uttarakhand

Email: prkshaiprince@gmail.com

Interests

Deep Learning

Generative AI

Natural Language Processing

Machine Learning

RL

Reinforcement Learning

Computer Vision

Meta Learning

Experience

Injala

November 2021 - Present

Senior Machine Learning Engineer

Solely led the successful design and implementation of an LLM-based chatbot using fine-tuning Llama2 for insurance policy use case. Additionally, I managed multiple user memory sessions with LangChain. Also developed a Python script to automate dataset creation for a reward-based chatbot system
I developed an end-to-end pipeline for Document AI within an average of 100 pages of PDFs. This pipeline includes dataset preparation, model training, testing, and deployment.
Used LayoutLM model, along with multiple graphics processing units (GPUs) and automatic mixed precision (AMP), to train a system that can summarize text.
Focused on using BERT for text classification and CNN models like VGG, ResNet, and GoogLeNet for pattern recognition. Additionally, I converted the model to the ONNX format.
Worked on developing scalable code that is suitable for production environments. This involved ensuring the code can handle high loads and traffic by conducting load testing on the API.
Deployed model as REST APIs into production server and merged multiple APIs into one.
Accelerated model inference time by leveraging batch sizes and multi-threading techniques.

Freelancer

july 2021 – Oct 2021

Machine Learning Engineer

Worked as a freelancer for a US based professional consultant. The project was funded by Bertrandt Inc.
Project on trajectory estimation for autonomous driving using Transformer neural network.
Developed scripts to filter data and generate features for trajectory prediction neural networks.
Trained model with different hyper-parameters, presented research paper for future trajectories of multiple agents.