Research Interests
I am interested in applying Machine Learning (ML), Recommendation Systems, Natural Language Processing (NLP), Causal Inference, Knowledge Graphs and Data Mining for health and well-being applications.
|
Skills
I have taken courses in Natural Language Processing, Information Retrieval, Artificial Intelligence, Machine Learning, Deep Learning, Algorithms and Data Structures, Probability and Statistics.
I am skilled in Python (TensorFlow, NumPy, SciPy, Matplotlib, Pandas), R, C, C++, SQL, Tableau, Flask, HTML, and CSS.
|
Open to Work and Work Authorization
I am looking for Summer 2024 internships as a Research Scientist Intern, Applied Scientist Intern, Data Scientist Intern, Software Development Intern.
I am open to remote work, can relocate anywhere in the U.S., have valid work authorization and do not need visa sponsorship. My expected graduation date is May 2025.
|
|
Towards Deep Personal Lifestyle Models Using Multimodal N-of-1 Data
Nitish Nagesh, Iman Azimi, Amir M. Rahmani, Ramesh Jain
In MultiMedia Modeling: 29th International Conference, MMM 2023, Bergen, Norway, January 9–12, 2023, Proceedings, Part I, pp. 589-600. Cham: Springer International Publishing, 2023
paper
|
|
World Food Atlas for Food Navigation
Ali Rostami, Nitish Nagesh, Amir M. Rahmani, Ramesh Jain
In Proceedings of the 7th International Workshop on Multimedia Assisted Dietary Management, pp. 39-47 , 2022
paper
|
|
University of California Irvine
Research Assistant
Irvine, CA
Analyze and visualize food, sleep, physical activity, nutrition logs over 3 years to provide N-of-1 context-aware personalized food recommendations using event mining, machine learning and Tableau.
Develop open-source World Food Atlas database to integrate multimodal food-related data across the globe.
Design novel data collection schema on top of Googles schema in collaboration with dietitians and physicians from Stanford University to standardize food-related dataset collection.
Develop personalized AI-driven applications to improve peoples mood through timely dietary interventions.
|
|
Qualcomm
Platform Intgeration Engineer
Austin, Texas, USA
Developed a Python tool to parse test data from 5000+ manufacturing logs of QAIC100 AI accelerator saving 3x cycle time.
Triaged and debugged failures in QAIC100 SDK using Linux scripting leading to a system-wide process change in the test methodology.
Involved in setting up proprietary server platforms in the corporate research and development lab for performance tests.
|
|
University of California San Diego
Researcher
San Diego, CA
Developed novel reliability-aware task allocation strategies for IoT networks using Python to reduce overall maintenance cost.
Built a real-world IoT mesh network communicating via MQTT and Wi-Fi to measure impact of resource constraints on reliability.
Achieved 90% accuracy while validating a reliability simulation framework for IoT networks using Python and C/C++.
Developed a remote monitoring tool to infer relationship between soil pH, soil conditions and ambient environment unlike traditional stand-alone systems
Effectuated targeted fertilizer application by calibrating a soil pH sensor with an average accuracy of 75%
Researched sensing techniques and machine learning approaches to measure and predict nitrate concentration in large scale deployments
Researched collaboratively on reproducibility for cyber-physical systems and IoT with case studies while referencing machine learning.
|
|
University of California Irvine
Teaching Assistant
Irvine, CA
Tutor and mentor 200+ students in the upper division course Critical Writing on Information Technology
Critically evaluate students elevator pitches, technical resumes, presentations and persuasive letters by providing constructive feedback preparing them to excel in corporate and academic roles
Led weekly discussion sessions for 20 students in Winter 2023 enabling mentees to draft better technical documents for proposing changes to existing communication and wellbeing platforms such as Gmail, Instagram, TikTok, Headspace etc
|
|
Natural Language Processing Implementation
Course Project:Natural Language Processing
Classified presidential candidate speeches via supervised and semi-supervised learning in Python/TensorFlow.
Built n-gram language models on the Brown, Gutenberg and Reuters corpuses. Analyzed in-domain and out-of-domain perplexities to compare language models and individual sentences.
Developed a part-of-speech (POS) and named entity recognition (NER) tagger for twitter data using Conditional Random Fields (CRF) and incorporated Viterbi algorithm to improve CRF accuracy.
Implemented top-K sampling, nucleus sampling, beam search decoding algorithms and evaluated summarization models qualitatively and quantitatively using Python/TensorFlow.
|
|
Slot Descriptions in Self-Attentive Dialogue State Tracking (DST)
Course Project:Natural Language Processing
Implemented full-shot and zero-shot dialogue state tracking on MultiWoz 2.1 dataset with 5 domains and 8438 dialogues using Python/TensorFlow to transfer knowledge from resource rich domains to unknown domains
Deployed BERT base model and evaluated accuracy for inserting slot descriptions in zero-shot and full-shot DST
|
|
Web Crawler and Search Engine Builder
Course Project:Information Retrieval
Crawled 50,000 URLs from ics.uci.edu domain using Python to find page similarity and subdomains
Built search engine using Flask, HTML, CSS to query and retrieve top twenty matches from crawled databases
|
|
Fashion MNIST Classification using Covolutional Neural Networks
Course Project:Artificial Intelligence
Classified fashion-MNIST dataset running convolutional neural networks (CNN) on Google Colab using Python
Achieved 95.88% training accuracy and 93% test accuracy after hyperparameter tuning and cross-validation
|
|
Reinforcement Learning and Machine Learning Algorithm Design
Course Project:Artificial Intelligence
Programmed reinforcement learning agent using Monte Carlo Tree Search in Python to solve Sokoban puzzle
Designed and implemented machine learning algorithms using kNN, Naïve Bayes classifiers, linear regression, cross-validation, logistic regression, shattering, nearest neighbor, decision trees, neural networks, and clustering
|
|
Interactive global energy consumption dashboard
Course Project: Renewable and Sustainable Energy
Developed first-of-its-kind energy parameter visualization platform for 200+ countries using Dash
Deployed scalable and globally accessible website using Heroku sourcing data from a structured SQL database using SQLite
Actualized user-friendly interface for parameters with customizable checkboxes and predictions using logistic regression in Python
|
|
Sensing and Actuation in Agriculture and Gardening Environmnets
Final Project: Introduction to Embedded Computing
Report /
Slides
Outperformed traditional sensing techniques with remote soil sensing and active real-time pest deterrence using
Linux, C/C++
Introduced predictive capabilities within 10% sensing range based on linear regression using the Scikit-learn
library in Python
Visualized soil vitals on an interactive online dashboard developed using HTML, CSS, Flask, and JavaScript
|
|
Food Waste Estimation using Received Signal Strength Indicator
Course Project: Embedded Computing and Communication
Report /
Slides
Attained 70% accuracy in determining an unknown amount of grocery waste using C/C++ and principles of RF
attenuation
Observed less than 25% standard deviation during prototype testing using received signal strength indicator
(RSSI) metric
Realized hands-off food waste estimation without modifying existing trash bin structure by simple retrofitted
add-ons
|
|
Real-time wireless ambient temperature sensing
Lab Project:Internet of Things
Developed wireless temperature sensing framework using a resistance temperature detector (RTD) sensor with less than 0.2 variation between sensed and actual values
Achieved 20% less external noise interference using a Sallen-Key low-pass filter in read-out circuit built using PSoC creator
Executed real-time secure communication with less than 5% latency using C/C++ with data encapsulation and visualization
|
|