Hey, I'm Kaif π
B. Tech CSE
I am currently pursuing a Bachelor of Technology degree in Computer Science and Engineering at D.Y. Patil International University, specializing in the fascinating realm of AI.
I have a strong interest in various research areas of machine learning and artificial intelligence, including Large Language Models, Human-Computer Interaction (HCI), Adversary Machine Learning, Interpretability.
In my studies and independent projects, I actively work with state-of-the-art AI technologies to expand my expertise. I am particularly passionate about conducting research at the intersection of data-driven NLP, TrustworthyML, computer vision, generative AI models.
I am always eager to tackle new challenges that combine my enthusiasm for technology with interdisciplinary collaboration. My goal is to conduct research that manifests theoretical ideas into practical tools with measurable impact. I believe bringing together diverse perspectives leads to truly transformative innovation.
Research Interests
Machine Learning; Natural Language Processing; AI Alignment; Large Language Models; Human-Computer Interaction (HCI); Adversary Machine Learning; Interpretability
Experience
Research Assistant
Indian School of Businessβ’ Developing a custom NLP pipeline with Vision OCR, Text Classification and Token Classification models using Open LLMs.
Research Intern
CISPA Helmholtz Center for Information Securityβ’ Topic of Research: "Privacy Preserving Generation with Large Language Models"
β’ Researching on Private Large Language Models using various Differential Privacy Frameworks.
β’ Analyzing several pre-training/fine-tuning methods and model behavior on various downstream tasks.
Research Assistant
Indian School of Businessβ’ Developed NLP Classifier Models on a daily basis, achieving high accuracy rates exceeding 95%, demonstrating strong data analysis and modeling skills.
β’ Expertly finetuned large transformer models, attaining precision and reliability, resulting in improved model performance and efficiency.
β’ Created Transformer models for Sequence Classification, Sequence Generation, and Token Classification on a daily basis, showcasing consistent innovation and adaptability.
β’ Specialized in developing customized Named Entity Recognition (NER) solutions, optimizing data extraction and information retrieval processes.
β’ Proficiently worked with Big Data to analyze patterns and successfully deployed developed models, contributing to data-driven insights and decision-making.
AWS & Machine Learning Intern
F13 Technologiesβ’ Worked with AWS EC2, EBS, and S3 to host and develop websites.
β’ Successfully migrated non-AWS web applications to AWS infrastructure.
β’ Developed a sophisticated Content Recommendation system using AWS Personalize.
β’ Gained hands-on experience with over 50 AWS services during the training period.
β’ Proficient in utilizing tools like AWS Cost Calculator to optimize cloud resources.
Research Intern
National Institute of Technology Patna (NITP)β’ Topic of research: Agile motion of quadrupedal locomotion using Quad-SDK
β’ Testing Deep Learning Algorithm efficiency for better obstacle detection and increasing accuracy by 25%.
β’ Using an efficient control scheme to increase in terrain mapping accuracy by 12.7% when using the right number of contours.
β’ Global Plannerβs code was optimized for a movement speed boost of 15%.
β’ Tools & Languages used: Linux, Git, ROS, Quad-SDK, Python, Catkin
Education
B. Tech β Computer Science & Engineering
D. Y. Patil International UniversityList of courses: Data Structures, Design & Analysis of Algorithms, Principles of Data Science, Intelligent Systems, Digital Signal Processing, Deep Neural Networks, High Performance Computing & Game Theory
Clubs & Co-Curriculars: TEDx Conferences, Google Development Students Club, Tech Cohorts and Hackathons.
Projects
Otaku Engine
Otaku Engine is a content recommendation system for anime enthusiasts (weebs). It leverages machine learning techniques to provide personalized anime recommendations based on user interactions and ratings.
Twitter Tweet Classification Model
Tweet Classification using Natural Language Processing. This NLP model classifies tweets as Disaster Tweets (1) or Non-Disaster Tweets (0). Model uses Logistic Regression(tf-idf), Naive Bayes(tf-idf), Logistic Regression to train the model and classify tweets.
iIMS β Intelligent Infrastructure Management System
Industry ProjectAn AI system aims to solve the obstacles faced by universities/organizations in making optimal use of infrastructure using OpenCV and ML Frameworks.
AI Maze Solver
AI based Maze Solver uses DFS or BFS Algorithm to encounter and solve maze's and helps you analyze which search problem has less path cost in similar scenarios.
Skills
Programming Languages
Python (w. RegEx), C++, C#, JavaScript, MySQL, MQL & Git
Tools & Technologies
Machine Learning (Transformers, TensorFlow & PyTorch), NLP, Computer Vision, Power BI & MongoDB
Cloud
AWS S3, EC2, Sagemaker, DynamoDB, Lambda and other popular AWS services.
Collaboration & Productivity
Asana, ClickUp, Notion, Jira, Miro & Deepnote
Additional Design Tools
Adobe XD, Figma, Blender, Cinema 4D & Unreal Engine