Priyansh Soni
Leveraging Data Science to enhance user experience
about me
I'm a professional data scientist with more than 2 years of experience advancing in Natural Language Processing and transforming personalised user experience via Recommendation Engines.
Natural Language Processing
From Classification and Named Entity Recognition, all the way to training and fine-tuning state-of-the-art LLMs, I have leveraged my expertise in developing advanced models for diverse Natural Language Processing tasks.
Recommendation Engines and Fueling Mechanism
With state-of-the-art models like Neural Collaborative Filtering (NCF), I have excelled in shaping content recommendations for a better user experience, alongside increasing user engagement by harnessing the power of modern Click Probability Prediction.
SKILLSET
The following points showcase my experience and skillset:
1. Natural Language Processing and Large Language Models:
Proficient in NLP techniques, specializing in the development and implementation of Large Language Models.
Experience in Training and Fine-tuning LLMs, Transformers, BERT for multiple NLP tasks such as Text Classification, NER, and Text Generation.
Proficient in utilizing various NLP libraries including Hugging Face, Gensim, Fasttext, and Tensorflow.2. Recommendation Engines and Fueling:
Specialized in designing and implementing advanced recommendation systems for personalized user experiences.
Developed models for click probability predictions, enhancing user engagement strategies.
Proficient in segmentation and clustering techniques, providing valuable insights for personalized data analysis and user experiences.
3. Tools, and Frameworks:
Proficient in using various tools and frameworks such as JupyterLab, TensorFlow, and MongoDB for comprehensive model development, analysis, and data storage.
Experience using Langchain and GPT3 LLMs for modelling and automation.
4. BigData and Cloud Platforms:
Skilled in leveraging Big Data technologies, particularly PySpark, for efficient data processing and analysis. Skilled in SparkML for model implementation and training
Proficient in utilizing cloud services, including AWS, Databricks, and Microsoft Azure Data Lake, for effective data handling and processing.
5. End-to-End Data Science Application Development:
Experienced in the complete life cycle of data science applications, from engineering and development to successful deployment.
6. Deployment and Infrastructure:
Expertise in deploying models using FastAPI, AWS Batch Job, SageMaker, AWS ECS, ensuring efficient and scalable solutions.
7. Continuous Maintenance and Monitoring:
Establishing continuous maintenance and monitoring setups for models, ensuring optimal performance over time.
RESEARCH PUBLICATIONS
1. ConvXGDFU -Ensemble Learning Techniques for Diabetic Foot Ulcer Detection - 4th IEEE ICAC3N-22, 2022
2. Study of Fake News Detection on Social Media Platforms - 6th RTEICT, 2021
Natural Language Processing
In the realm of Natural Language Processing (NLP), my expertise extends across a diverse spectrum of techniques, methodologies, and tools, fostering a comprehensive understanding of language-driven data. Proficient in NLP techniques, I have specialized in the development and deployment of Large Language Models (LLMs) that form the cornerstone of various applications. Leveraging advanced models such as Transformers, BERT, and other state-of-the-art frameworks, I have actively engaged in training and fine-tuning processes, enabling these models to excel in multiple NLP tasks. This encompasses tasks ranging from Text Classification, where models discern and categorize textual content, to Named Entity Recognition (NER), identifying and classifying entities within the text, and Text Generation, which involves the creation of coherent and contextually relevant textual content.My skill set extends to the utilization of prominent NLP libraries, including Hugging Face, Gensim, Fasttext, and Tensorflow. Harnessing the power of these libraries, I have implemented robust solutions for complex NLP challenges. Whether it be the seamless integration of pre-trained models from Hugging Face or the development of custom solutions using Tensorflow, these libraries have been instrumental in crafting innovative and efficient NLP applications. The synergy between theoretical knowledge, practical experience, and a diverse set of libraries ensures a holistic approach to NLP, contributing to a rich tapestry of language-based data applications.
Recommendation Engines and Fueling Mechanism
Within the domain of personalization strategies, my focus lies in the design and implementation of advanced recommendation systems tailored for delivering personalized user experiences. With a keen understanding of user behavior and preferences, I have developed and deployed recommendation models that go beyond conventional approaches. These models incorporate sophisticated algorithms and methodologies to understand user interactions, providing recommendations that align seamlessly with individual preferences.One significant area of expertise is the development of models for click probability predictions, a critical component in enhancing user engagement strategies. By leveraging machine learning techniques, these models accurately predict the likelihood of user clicks on various elements, enabling the implementation of targeted and personalized content delivery. Additionally, my proficiency extends to segmentation and clustering techniques, providing valuable insights through data analysis to further refine and customize user experiences. This involves categorizing users based on behaviour patterns, allowing for the creation of user segments and the delivery of highly targeted and relevant content.