top of page
Rivers__1_-removebg-preview.png

SAMEER MAURYA

ABOUT

With over seven years of experience, I am a seasoned Data Scientist specializing in Generative AI project implementation, Large Language Models (LLMs) for financial services, Computer Vision, and Natural Language Processing (NLP). I have handson expertise in developing, validating, and deploying LLM-powered financial applications, ensuring compliance with AI regulations and model risk management frameworks. My technical proficiency encompasses Python and advanced deeplearning frameworks. I have led significant computer vision and deep learning projects, particularly tailored for edge computing environments. My portfolio includes deploying various deep-learning solutions on platforms such as AWS, Heroku, and Azure, providing a comprehensive understanding of the entire machine learning project lifecycle. Additionally, I possess expertise in Model Risk Management and stakeholder communication. I am also skilled in Responsible AI and cybersecurity, ensuring the development and deployment of secure, ethical, and compliant AI solutions that meet industry standards.

IMG_20190210_144625_Bokeh_edited.jpg
EXPERIENCE
720px-Bank_of_America_logo.svg.png
720px-Bank_of_America_logo.svg.png

  • Generative AI & LLM Validation: Developed and validated LLM solutions (BART, GPT, Llama) for financial services, ensuring adherence to Model Risk Management (MRM) and AI regulatory frameworks. -LLM Implementation: Led a BART-based summarization project that reduced manual effort by 90% and streamlined automated risk assessments and document summarization.

  • Information Security Modeling (GIS): Engineered and validated Identity Threat Detection & Response (ITDR) models, specifically focusing on LDAP Reconnaissance, Account Takeover (ATO) detection, and Algorithmic Identity Linking. -Security & Robustness Testing: Executed independent validation tests focusing on adversarial robustness, explainability, and vulnerability assessments (e.g., prompt injection) for LLMs and security models.

vf.png

  • Built features for the bot platform, including Named Entity Recognition (NER) and intent detection (SURBO).

  • Worked closely with the Product team to create different ML features in the chatbot platform (SURBO)

  • developed Optical Character Recognition (OCR) as a service for PAN, ADHAAR, and Structured Receipts so that it can be utilised by a variety of Bot products used by the organisation.

  • We developed a unified machine learning application programming interface (API) that all our Chatbot products and the rest of the organisation can use.

vf.png
chamko_white.png
chamko_white.png

  • I developed a recommendation system and established 15 unique data points to be associated with each influencer in order to assist me in selecting the most effective influencer.

  • I created a model to predict ages and genders with the help of Keras. First, I labelled the data with pre-trained weights, and then I trained the model using the data that the user provided.

  • Interacted with various company executives to showcase the data points that had been gathered during the campaigns they were running.

cropped-Synergylabs-logo-with-name1-279x77.png

  • I led the research team as we designed a graphical user interface (GUI)-based image preprocessing solution for quality control and quality assurance automation that could be deployed on edge (using NVIDIA Jetson nano) and combined with robots manufactured by Universal Robotics.

  • We were able to build an automated attendance system based on facial recognition that has a 97.8 per cent accuracy rate after just needing a single photo of the user to train it with.

  • Through the implementation of a dense captioning research paper, we were able to construct an image-captioned system.

cropped-Synergylabs-logo-with-name1-279x77.png

REACH OUT TO ME

bottom of page