# Github Information: Hello, I’m Vrinda :) I write code and enjoy research problems. I have a bachelor's in Computer Science too! Here’s what I’m up to: Machine Learning Engineer @ Gloroots Developing recommendation solutions for the HRTech space. Research @ Boston University (BITLAB) Exploring NLP and AI from a supply chain lens. Current Side Project: Replicating various adversarial attacks (slowly but surely!). Research Interests I’m really interested in: Differential Privacy Federated Learning Algorithmic Fairness Always up for a chat or collaboration on these topics! Hobbies When I’m not working on tech projects, I’m an avid reader. I also paint. Currently reading: Rhythm of War by Brandon Sanderson – book 4 of Stormlight Archive (10/10, would recommend), doing a reread before digging into the latest Wind and Truth! # Resume: VRINDA KOHLI (+91) 8091701861 ⋄Email ⋄LinkedIn ⋄GitHub ⋄Website EDUCATION Bachelor of Technology , Manipal University Jaipur 2020 - 2024 Computer Science Engineering CGPA : 9.26 Awards: 4x Dean’s List, Award for Academic Excellence Minor: Data Science EXPERIENCE Machine Learning Engineer Jan 2024 - Present Gloroots Bengaluru, Karnataka •Orchestrated end-to-end enrichment pipeline for technical and sales teams, elevating data coverage by more than 90%. •Leveraged open source LLMs (Mistral, LLaMA) for building 3 in-house applications, and designed custom AI agents. •Contributing to development of embedding based job recommender engines, focusing on low latency reranking algorithms. •Facilitating seamless API creation and integrations within the Django, MongoDB and PostgreSQL based product workflow. Research Assistant August 2023 - Jan 2024 BITLab, Boston University Remote •Developed a scalable tool for scraping product details from Amazon, successfully capturing 15+ attributes for over 30,000 products. •Conducted sentiment analysis and experimentation on customer insights using scraped data to identify correlations between sales and the green attributes of skincare products. •Contributed to a LangChain-based multi-agent environment for experimenting with inter-agent interactions, utilizing tools like SerpAPI and OpenAI models. Research Intern May 2023 - August 2023 Delhi Technological University New Delhi •Used transfer learning and transformer models for flash flood detection using geospatial raster imaging, improving perfor- mance by 8%. •Implemented and validated 7 recent research papers on Deep Learning aided Metaphor Detection, including transformer and RNN based Siamese approaches. •Achieved 2% average accuracy improvement over existing methods during experimentation. Research and Development Intern July 2022 - Nov 2022 Trish-i, IIT Mandi Catalyst Mandi, Himachal Pradesh •Developed end-to-end pipeline for classification task: from data collection and labeling to deployment via Flask. •Trained VGG and YOLO based CNN architectures to predict 5 bone health conditions from X-Ray images with upto 94% accuracy using Tensorflow and Keras. •Finetuned aforementioned models, leading to an 11% performance spike. PUBLICATIONS •Cracking the Figurative Code: A Survey of Metaphor Detection Techniques Presented, ADCIS’23 •Finding GAIA (Generative AI Art) Online Under Review •Observing the Privacy-Utility Tradeoff in Differentially Private Medical Text Classifiers Under Review PROJECTS AI Generated Art Detection :JPEG-Compression Agnostic Detection of Generative Art on Social Media. •Designed and implemented a lightweight architecture for detecting StyleGAN2-ADA and Stable Diffusion generated images using feature engineering and XGBoost ensembles. •Achieved over 95% accuracy while using less than 20% of standard training data. GPT Implementation :Generatively Pretrained Transformer for generating Shakespearean-style quotes. GitHub •Developed a GPT model architecture and its components(tokenizers, multihead attention) from scratch using Python. PATE Implementation :Semi-supervised Knowledge Transfer for DL from Private Training Data. GitHub •Implemented the seminal PATE paper for Differentially Private training using PyTorch and PySyft on the MNIST dataset. SKILLS Languages Python, C/C++, JavaScript, SQL Frameworks PyTorch, PyTorch Lightning, Langchain, Flask, Django, FastAPI Tools Git, Linux, Figma, Google Cloud Platform (GCP)