๐Ÿš€ For my latest work on building with weird yet usable AI & safety/privacy research: ๐Ÿ“š Read My AI Blog. The below is my academic portfolio!

About Me

A Private n' Fair AI Enthusiast (Check out my Blogs!)

I am Aman Priyanshu, a Masters Student at Carnegie Mellon University, with a specialization in Privacy Engineering. My research interests include Privacy Preserving Machine Learning, Explainable AI, Fairness, and AI for Social Good. Currently I am an AI Security Researcher at Robust Intelligence and one of my works on bypassing Meta's AI safety system has been attributed across some news articles (1, 2, 3, 4, 5). I have recently focused on areas such as Prompt Engineering, Reinforcement Learning, Causal Inference, and Bias Mitigation. During my undergraduate studies, I was recognized as an AAAI Undergraduate Scholar (2023) for my contributions in the field of Fairness and Privacy. Additionally, I have had the opportunity to gain practical experience through internships, including working as a Privacy Engineer Intern at Eder Labs R&D Private Limited, a MITACS Research Intern at Concordia University, and as an Undergraduate Research Assistant at Manipal Institute of Technology.

I have also had the opportunity to have several publications in the field of machine intelligence and security and have won awards for my projects. My explorations in technology are also extended to participating in hackathons, where I've applied my research to developing applications aimed at social good. Check out my AI blog and also a pick of my favorite models

Aman Priyanshu

Curriculum Vitae

           

Experience

Education

Publications

2024 | 2023 | 2022 | 2021 | 2020


2024

  1. Through the Lens of {LLMs}: Unveiling Differential Privacy Challenges    Link  
    Aman Priyanshu, Yash Maurya, Vy Tran, Suriya Ganesh Ayyamperumal
    Accepted at 2024 USENIX Conference on Privacy Engineering Practice and Respect

2023

  1. Guarding Your Social Circle: Strategies to Protect Key Connections and Edge Importance    Link  
    Nisha P Shetty, Balachandra Muniyal, Akshat Dokania, Sohom Datta, Manas Subramanyam Gandluri, Leander Melroy Maben, Aman Priyanshu
    Accepted at Security and Communication Networks
  2. FedBully: A Cross-Device Federated Approach for Privacy Enabled Cyber Bullying Detection using Sentence Encoders    Link  
    Nisha P Shetty, Balachandra Muniyal, Aman Priyanshu, and Vedant Rishi Das
    Accepted at Journal of Cyber Security and Mobility
  3. Are Chatbots Ready for Privacy-Sensitive Applications? An Investigation into Input Regurgitation and Prompt-Induced Sanitization    Link  
    Aman Priyanshu, Supriti Vijay, Ayush Kumar, Rakshit Naidu, and Fatemehsadat Mireshghallah
    ArXiv Submission

2022

  1. #maskUp: Selective Attribute Encryption for Sensitive Vocalization for English language on Social Media Platforms    Link  
    Aman Priyanshu and Supriti Vijay
    Presented at the Research & Reports Track at #ShowYourSkill organized by Coursera
  2. NERDA-Con: Extending NER models for Continual Learning - Integrating Distinct Tasks and Updating Distribution Shifts    Link  
    Supriti Vijay and Aman Priyanshu
    Accepted at the Updatable Machine Learning Workshop, ICML 2022
  3. ARLIF-IDS--Attention augmented Real-Time Isolation Forest Intrusion Detection System    Link  
    Aman Priyanshu, Sarthak Shastri, and Sai Sravan Medicherla
    Accepted at the Poster session at the 43rd IEEE Symposium on Security and Privacy
  4. Finding an elite feature for (D)DoS fast detection-Mixed methods research    Link  
    Josy Elsa Varghese, Balachandra Muniyal, and Aman Priyanshu
    Journal: Computers & Electrical Engineering, Volume: 98, Pages: 107705

2021

  1. Efficient Hyperparameter Optimization for Differentially Private Deep Learning    Link  
    Aman Priyanshu, Rakshit Naidu, Fatemehsadat Mireshghallah, Mohammad Malekzadeh
    Accepted at the Privacy Preserving Machine Learning Workshop, ACM CCS 2021
  2. Something Something Hota Hai!" An Explainable Approach towards Sentiment Analysis on Indian Code-Mixed Data    Link  
    Aman Priyanshu, Aleti Vardhan, Sudarshan Sivakumar, Supriti Vijay, and Nipuna Chhabra
    Accepted at Workshop on Noisy User-generated Text (W-NUT), EMNLP 2021
  3. When Differential Privacy Meets Interpretability: A Case Study    Link  
    Rakshit Naidu, Aman Priyanshu, Aadith Kumar, Sasikanth Kotti, Haofan Wang, and Fatemehsadat Mireshghallah
    Accepted at the Responsible Computer Vision Workshop, CVPR 2021 and Privacy Preserving Machine Learning Workshop, ACM CCS 2021
  4. Continual Distributed Learning for Crisis Management    Link  
    Aman Priyanshu, Mudit Sinha, and Shreyans Mehta
    Accepted at the 3rd Workshop on Continual and Multimodal Learning for Internet of Things, IJCAI 2021
  5. FedPandemic: A Cross-Device Federated Learning Approach Towards Elementary Prognosis of Diseases During a Pandemic    Link  
    Aman Priyanshu and Rakshit Naidu
    Accepted at the Machine Learning for Preventing and Combating Pandemics and the Distributed and Private Machine Learning Workshops, ICLR 2021

2020

  1. Stance Classification with Improved Elementary Classifiers Using Lemmatization (Grand Challenge)    Link  
    Aman Priyanshu, Vedant Rishi Das, Shashank Rajiv Moghe, Harsh Rathod, Sai Sravan Medicherla, Mini Shail Chhabra, and Sarthak Shastri
    Accepted at 2020 IEEE Sixth International Conference on Multimedia Big Data (BigMM)

Relevant Projects

  • Jun-23: ProTaska-GPT
    Specify your dataset of choice, and ProTaska-GPT will understand the dataset with tasks, tutorials, and actionable insights for it. Accelerate your data science journey with ease and efficiency! (Meant for people starting their journey into Data Science.)
  • Oct-22: AdaptKeyBERT
    Built a python library, integrating semi-supervised attention for creating a few-shot & zero-shot domain adaptation technique for keyphrase extraction.
  • May-22: NERDA-Con
    NERDA-Con is a python package, a pipeline for training Named Entity Recognition (NER) with Large Language Models bases by incorporating the concept of Elastic Weight Consolidation (EWC) into the NER fine-tuning NERDA pipeline
  • Jun-22: DP-SDV
    Creating a Differential Privacy securing Synthetic Data Generation for tabular, relational and time series data.
  • Aug-21: DP-HyperparamTuning
    DP-HyperparamTuning offers an array of tools for fast and easy hypertuning of various hyperparameters for the DP-SGD algorithm. We proposed a novel, customizable reward function that allows users to define a single objective function for establishing their desired privacy-utility tradeoff.
  • Aug-21: Hexa Lite
    Created an unsupervised machine learning to extract contextually similar texts. The project was used in indexing Academic Literature, Law Precedents, and Financial Records. The project won Code Innovation Series - a Hackathon in association with GitHub.
  • Jul-21: Augmented Face Detection API for Professional Image Approval
    The app performs obstruction detection, spoof detection, blur detection and environment approval. Utilized Deep Neural Networks and Genetic Algorithms to achieve these goals in low computational time. The project won 1st place in HackRx 2.0 by Bajaj Finserv.
  • May-21: DeCrise
    DeCrise is an online platform that acts as an aggregator for public support/utility services which uses continual-federated-learning to create a quick response information retrieval system during a natural disaster. The project won 1st place in The ACM UCM Datathon.
  • Apr-21: Voix
    A social-media platform employing machine learning and differential privacy to promote civic engagement while protecting user-privacy. The project won under the Community & Civic Engagement for UC Berkeley's CalHacks Hackathon

Achievements

  • 2024 March: Spark Grant Winner - NOVA Hacks
    Won the Spark Grant for our app that enhances speech for non-native English speakers, employing prompt-engineering function-calling (OpenAI GPT4/3.5) and Speech-to-Text (OpenAI Whisper), with features like audio-segmentation, speaker-recognition, and diarization. The app incrementally learns from daily conversations to develop lessons, boosting confidence, vocabulary, and local phrases/colloquialisms, while ensuring privacy through audio-based Notice & Choice.
  • 2023 September: Space Theme Category Winner - HackCMU
    Won the Space-Themed track with our space trash collection project using Pareto optimization to balance time, fuel requirements, satellite movements, planetary alignment, and the trajectory of trash collectors for predicting monetary incentives.
  • 2023 February: Research & Travel Grant - AAAI Undergraduate Consortium Scholar
    Selected as one of twelve individuals for the AAAI UC program, recognizing my research on Privacy and Fairness.
  • 2022 June: Second Runners-Up - ShowYourSkill (Coursera)
    Came second runners-up in #ShowYourSkill where we participated in the Research & Reports Track and creating a NLP augmented Machine Learning Application for women safety.
  • 2021 September: Runners-Up - BobHacks 2021 (MetaBob API)
    Came runners-up in BobHacks where we built a pattern recognition API built on top of the MetaBob API. The API is able to assist users in tracking common errors and delivers pattern recognition on the MetaBob API.
  • 2021 August: First Prize - Code Innovation Series - associated with GitHub
    Innovation Series Hackathon was the hackathon organized by by Manipal Institute of Technology. Employed Document-Embedding for measuring contextual similarity between multiple pages and given search-queries.
  • 2021 July: First Prize - HackRx by Bajaj Finserv
    HackRx is the Annual Hackathon hosted by Bajaj Finserv. Used Deep Learning and Classical Image processing to achieve a face verification and profile-rank estimation task. The methodology out-performed classic Deep Learning methods. Created an API for the same.
  • 2021 May: First Prize - ACM UCM Datathon - UC Merced
    Won the ACM UCM Datathon, built DeCrise, an online platform that acts as an aggregator for public support/utility services for fast-response during a major crisis or disaster.
  • 2021 April: Runners-Up - Paper Presentation - IEEE SBM Manipal
    Presented a preliminary investigation of Federated Learning integrated with Continual Learning for Crisis Management.
  • 2021 April: First Prize - Community & Civic Engagement track of CalHacks Hackathon
    Won under the Community & Civic Engagement track of CalHacks Hackathon organized by UC Berkeley. Built Voix, an anonymous platform for uplifting communities and promoting civic participation. It is a social media platform that utilizes privacy-enabled machine learning to recover ideas affecting communities and bring them to the top of our platform while conserving user identity.
  • 2020 September: Runners-Up - Furniture Identification - IECSE x VISION
    Employed skip-connections to generate high-performance model for furniture identification.
  • 2020 August: Runners-Up - IEEE BigMM Data Challenge - IEEE Grand-Challenge
    Came runners-up in IEEE Grand-Challenge for harassment detection on tweets. Utilized Elementary Classifiers for Sentiment Analysis. The team was invited to present a paper in IEEE Sixth International Conference on Multimedia Big Data (BigMM).
  • 2020 January: Scholarship Recipient - Intel Labs
    Honored to be selected as one of the recipients of the Intel Edge AI Scholarship Program. Learnt about Machine Learning Implementation on the Edge.