Sree Bhargavi balija

Actively looking for summer intern positions as a ML/AI Engineer.

I am a graduate student at The University of California - San Diego where I'm pursuing my master's, majoring in Machine Learning & Data Science. I have worked on projects that involve reinforcement learning, Rest Api integrations, docker, time series analysis, computer vision and recommendation systems, and have experience with Software technologies such as Apache Spark, Apache Flink and Docker. I am also familiar with tools for MLOps, such as Prefect and MLflow, and have experience working with cross-functional teams to integrate Api's into business processes.

Prior to joining graduate school, I have worked as Software Engineer for 2+ years at ServiceNow where I delved the ITSM virtual agent topics on the NOW-bot for managing and operating Citrix virtual systems and Designed the success dashboard which provides a prebuilt analytics to demonstrate the actual business value achieved through ServiceNow products.

Before working full-time I was pursing my Bachelors at Indian Insitute of Technology, Hyderabad with majors in civil engineering and with specific interest in Machine learing.

Feel free to check out my Resume and drop me an e-mail if you want to chat with me!

~ Email | Resume | Github | LinkedIn | Tech trends ~

Sep '22	Started Fall 2022 Quarter - U.C. San Diego!
June '22	Completed my 2 years working as a Software engineer at ServiceNow.
June '20	Completed my Bachelors.

University of California - San Diego

Master of Science | Machine Learning & Data Science
Sep '22 - Present
Relevant Coursework:
ECE-285: Deep generative models • CSE-251: AI learning algorithms • ECE-276A: Sensing and Estimation • ECE-250: Random Processes • CSE-258: Recommender Systems and Data Mining • CSE-257: Search and optimization • ECE-176: Linear algebra and its applications • ECE-176: Introduction into the deep learning and applications • ECE-143: Programming for data analysis • ECE-271A: Statistical Learning

Indian Institute of Technology, Hyderabad

Bachelor of Technology | Engineering
Jul '16 - May '20
Relevant Coursework:
CS-2443: Algorithms • CS-3550: DBMS • CS-1353: Introduction to the data structures • CS-2233: Data structures • EE-5602: Probabilistic Graphical Models • AI 5001: Artificial intelligence

Software Engineer | ServiceNow
July '18 - Aug '21

• Worked on integrating multiple rest api’s with ITSM workflows for adding virtual bot capabilities like Citrix cloud virtual systems access, manage meetings and request item flow, also integrated the topics with an NLU model for intelligent conversation flow
• Designed and developed the success dashboard which provides a prebuilt analytics to demonstrate the actual business value achieved through ServiceNow products.
• Streamlined the java code to demonstrate WebRTC screen share between Androids or Desktop browsers.

Summer Intern | myhome
May '19 - July '19

• Researched on various new technologies like optimization of scrap iron, digital elevation models in the industry and performed cost-benefit analysis for checking the feasibility of the product.

	Developing Communication efficient asynchronous peer to peer federated LLMs \| UCSD March 2024 • Introduced a novel, fully decentralized federated learning framework specifically tailored for Large Language Models (LLMs), leveraging blockchain-federated LLM (BC-FL) algorithms to enhance the balance between latency and accuracy in decentralized learning environments.
	• CPTQuant - A Novel Mixed Precision Quantization Techniques for Large Language Models \| UCSD June 2024 • Efficiently balances precision and computational efficiency for LLMs. Reduces model size while maintaining high accuracy. Optimized for deployment on GPUs and edge devices.
	FedNAM+: Executing Interpretability Analysis using Novel Conformal Predictions method \| UCSD March 2024 • Enhances model transparency through conformal prediction-based analysis. Provides reliable uncertainty estimates for federated learning models. Improves decision-making by quantifying prediction confidence.
	Interpretable federated learning through Neural additive models \| UCSD Dec 2024 • Built novel Federated Neural Additive Models architecture (FNAMs) which combine the interpretability of Neural Additive Models (NAMs) with the privacy-preserving attributes of federated learning, enabling decentralized learning across multiple devices or servers while ensuring data privacy.

	Reddit-sentimental-analysis Sentimental search, Pandas, Sentence transformers, TF-IDF, Unigrams, Jupyter Notebook \| Sep. 2022 - Dec. 2022 [code][report] • In this project, we have built a sentence transformer model for classifying the Reddit titles with the subreddit classes, we have used all-mpnet-base-v2 and multi-qa-mpnet-base-dot-v1 semantic search models to map reddit titles with 473-dimensional dense subreddit vector spaces. • We have also built word2vec, unigram, bigram and similarities models from the most correlated variables which we got from exploratory data anaysis • Further, we have evaluated the models using accuracy and precision, recall metrics. Accuracy was found to be 60% for sentence transformer models
	Summarizing LinkedIn feedposts using ChatGPT Python, Jupyter Notebook, LinkedIn, OpenAI api's \| Jan 2023 [code] • In this project, I have used chatGpt Api to summarize my LinkedIn feed posts • I have used LinkedIn and OpenAI's REST APIs to get started. After fetching the posts on my feed, I have passed them over to ChatGPT as a prompt.
	Tictactoe-reinforcement-learning Python, Pandas, Epsilon greedy techniques, Jupyter Notebook \| Jan. 2020 - May. 2020 [code] • Build an RL agent that learns the game by Q-Learning by choosing the hyperparameters such as epsilon (decay rate), learning-rate, discount factor. We have trained the model iteratively to obtain a good combination of hyperparameters. • Streamlined the python code for implementing a TIC TAC TOE playing algorithm using epsilon greedy method, Reinforcement learning technique.
	Apache spark Twitter data structured streaming PyTorch, Pandas, Sentimental analysis, Jupyter Notebook \| Sep. 2022 - Dec. 2022 [code][report] • Initially authentication operations and keys were obtained from twitter api and Python module "Tweepy" • A later stream listener named "Twitter data" was created to generate data for the kafka topic "Global warming". This new method was included in Twitter data object using Affinn module for calculating the sentimental value of tweet • Further the streaming data is converted into the structured data and placed in sql table named "SQldata" which has two columns "text" and "senti_val" • Pyspart.sql functions are used to calculate the average of sentimental values of the senti_val column, function fun is added to categorize the tweet to positive, negative or neutral based on the score.
	Google-youtube-data-Api Pandas, Youtube api, Jupyter Notebook \| Dec. 2022 [code] • YouTube API is an application programming interface that allows you to embed videos, curate playlists, and offer other YouTube functionalities on your website • I have added a python script to get all the video titles of particular channel using youtube api and then developed a plot between count of videos to the uploaded hour of a day.
	Phone-location-detection Tensorflow, Keras, Pycharm \| Sep. 2022 - Dec. 2022 [code][report] • In this project, my task is to implement a prototype of a visual object detection system to find a location of a phone dropped on the floor from a single RGB camera image • we have used a deep neural network architecture for the phone location detection model. Initially multiple images have been trained using VGG16 Architecture and then tested with images of 10% dataset. Accuracy of model is observed to be 70%.
	Deep learning project in VIGIL lab\| Prof.C Krishna Mohan, IITH PyTorch, scikit-learn, Pandas, Plotly-express, Jupyter Notebook \| Jan. 2022 - Mar. 2022 [code] • Worked on real-time object detection of videos using OpenCV Deep neural network module for traffic videos generated from CCTV footage • Researched various deep learning techniques for crowd density estimation. • Built deep neural network model for the gap acceptance of driving vehicles at Unsignalized intersections, class balance analysis and feature selection.
	Dietary Analysis Pandas, Numpy, Plotly, Jupyter Notebook \| Sep. 2022 - Dec. 2022 [code][slides] • Project focused on detailed analysis of dietary intake and dental health to help users beware of potential health risks before purchasing a product • Intially data was preprocessed using 0-1 scaling and PCA for numerical data and one hot encoding for Categorical data, models were built on top the data (Train-Test Split : 80%-20%) using random forest, SVM and Ada boost • Highest accuracy of 70% was found for ada boost algorithm whereas svm model shown 68% and Random forest shown 66% model accuracy.
	Docker image Docker, Visual studio \| Dec 2022 [code] • Developed docker image for graph algotihms and published to docker hub.

	Teaching Assistant \| Website management Jul '18 - Apr '19 Working under the supervision of Prof. Surendranath Somala at the IIT Hyderabad, for managing the iith website.
	Teaching Assistant \| Introductory course in chemistry Jul '18 - Apr '19 Working under the Prof. Surajith maity at the IIT Hyderabad • Mentored students in understanding the concepts of introductory physics and chemistry courses, helped undergraduate students in their course projects • Assisted the course instructor in grading weekly assignments.

Class representative | Class of 2020
Jul '18 - Apr '19

• Worked as a class representative for 2018 batch, acted as a liaison between the class and professor.
• Acted as a prime official channel of communication between teacher and rest of the class, for all monitoring formalities.
• Dealing with matters, which benefits to the class like shifting of class hours, tutorials, exam schedule, alteration of marks distribution etc.

Nss Volunteer | Vidhyadhaan, Plantation activities
Jul '16 - Apr '20

• Worked towards educating underprivileged children and rural minorities promoting sustainable development and associated practices that can be inculcated in daily life.
• Was an active participant in plantation and cloth donation activities

This template is a modification to Jon Barron's website. Find the source code to my website here.