Apoorvi Dwivedi

JavelinI

AI Solutions Engineer Intern • 10/2023 - Present

• Engaged in fine-tuning Language Models (LLMs) specifically tailored for financial applications, focusing on tasks such as analysis, report generation, and information retrieval.
• Implemented and optimized Retrieval-Augmented Generation (RAG) techniques to improve natural language understanding and information extraction in financial contexts.
• Proficient in working and fine-tuning GPT, Mistral and various open-source LLMs to enhance their capabilities for financial use cases.
• Supported the design and deployment efforts for cutting-edge AI solutions during the internship, playing a role in the development of a platform that redefines private capital data analysis.

HSBC Technology India

Software Engineer • 09/2019 – 09/2022 (3yOE)

• Developed and deployed over 40 RESTful API services using SpringBoot on Pivotal Cloud Foundry and AWS platforms, effectively enabling seamless integration and communication between systems.Utilized Kubernetes for container orchestration and leveraged AWS services such as Amazon EC2, and Amazon S3 for scalable and reliable infrastructure.
• Demonstrated leadership by overseeing the complete API lifecycle development and successfully implementing automated tests, leading to a notable 40% increase in code quality and overall system reliability.
• Proactively implemented measures to prevent API incidents, resulting in a significant 30% decrease in reported incidents and ensuring uninterrupted service delivery to users.
• Actively participated in proof-of-concept (POC) projects aimed at automating and innovating daily operational tasks, driving efficiency and productivity within the organization.

Queen Mary University of London

MSc Artificial Intelligence • 2022 — 2023

Distinction

Deep Learning for Audio and Music, Applied Statistics, AI in gaming, Interactive Agents, Computational Creativity, Machine Learning, Reinforcement Learning

Galgotias College of Engineering and Technology

Bachelors in Computer Science • 2015 — 2019

Distinction

Data Structures and Algorithms, Data Mining & Warehousing, Computer Networks, Pattern Recognition, Digitial Image processing, Distributed Systems, Operating Systems and Computer Graphics

Enhancing Music Emotion Recognition in Classical Pieces through Fine-Tuning

• Developed a novel transfer learning approach to attention fine-tuning for Emotion Recognition in Classical music pieces.
• The project aimed to refine the models ability to discern and categorize complex emotional states embedded within the musical fabric.

Audio Sound Classification using CNNs

• Developed a sound classification system using deep learning techniques and CNN. Employed spectrogram representations to train a highly accurate model capable of effectively classifying a wide range of audio sounds, including speech, music genres, and environmental noises.
• Demonstrated expertise in TensorFlow by utilizing its powerful capabilities for model development, optimization, and deployment in the sound classification project.
• This project uses advanced audio processing techniques, such as feature extraction and data augmentation, to enhance the performance of the model and achieve superior classification accuracy.

MusicGAN for Generating Lofi Music using Emotions in a Poetry

• Developed a MusicGAN model that employs advanced deep learning techniques, including recurrent neural networks (RNNs) and attention mechanisms, combined with natural language processing to generate emotionally resonant Lofi music based on textual input.
• This project showcased a unique fusion of cutting-edge technologies and fine-tuning complex Deep Learning architectures to capture the subtle nuances of music composition.

Sentiment-Analysis-imdb with Fine-Tuned BERT

• Developed and implemented a sentiment analysis project using the BERT architecture, fine-tuned on a sentiment analysis dataset. The goal of the project was to classify text into positive or negative sentiment categories.
• The trained sentiment analysis model is hosted on the Hugging Face Model Hub, ensuring easy access and integration for future projects and collaborations.
• Designed and deployed a user-friendly Streamlit web application, allowing users to perform sentiment analysis interactively.

Image Classification on MNIST Dataset

• Developed and implemented a robust PyTorch-based deep learning model with a CNN architecture to accurately classify handwritten digits from the MNIST dataset.
• Demonstrated proficiency in Python, deep learning principles, and the PyTorch framework.

Reinforcement Learning to Play Tabletop Games

• Applied reinforcement learning algorithms to train an agent capable of playing tabletop games.
• Implemented algorithms like Q-learning or Deep Q-Networks (DQN) to enable the agent to learn optimal strategies.

Galgotias College of Engineering and Technology

Chief Secretary • 2018 — 2019

As the Chief Secretary at Galgotias College of Engineering and Technology, I undertook the crucial responsibility of efficiently managing and organizing a wide array of cultural and technical events spanning the entire college. Through my diligent coordination and planning, I ensured the smooth execution of these events, fostering a vibrant and enriching campus atmosphere.

Galgotias Collge of Engineering and Technology

Club Head (Creative Team) • 2017 — 2018

As the Club Head of the Creative Team at Galgotias College of Engineering and Technology, I successfully led a team of 20+ individuals in organizing and managing engaging club activities. I played a pivotal role in conceptualizing and executing events while overseeing the creation of captivating promotional content.

Department of Computer Science and Engineering

General Secretary Extreme Team, Departmental Club CSE • 2017 — 2018

As the General Secretary of the Extreme Team, a departmental club under the Department of Computer Science and Engineering, I played a crucial role in managing and organizing various cultural and technical events. I excelled at promoting effective communication between the student body and the departments management, ensuring that student concerns were addressed promptly and efficiently. Through my dedicated efforts, I contributed to fostering a vibrant and engaging environment within the CSE Department, encouraging active participation and collaboration among students and faculty.