Juhi Mittal

Education

• Master of Science in Computer Science and Engineering, University of California, San Diego, 2022-2024
• M.Sc. (Hons.) Economics and B.E. (Hons.) Computer Science, Birla Institute of Technology and Science (BITS) Pilani - India, 2016-2021 [GPA: 9.48/10]

Work experience

• September 2021 - Present: Visiting Associate - Researcher
  • Location: World Well Being Project, University of Pennsylvania
  • Duties Included:
    • Built computational models using ML and NLP techniques for studying phasal language change and social listening for mental health during COVID.
    • Detecting vaccine related misinformation for 1billion+ tweets using transfer learning and graph based BERT models.
    • Lead the technical analysis of the projects in collaborations with World Bank and PennMedicine by handling the implementation and presenting findings to stakeholders and quickly picking the domain knowledge.
    • Implemented spark-based mapper to map the user location to respective admin levels and wrote the Twitter and Reddit API puller used by the group.
    • Tools used: Spark, Hadoop, Natural Language Processing, Twitter API, Reddit API, Pandas, Keras, SQL
• January 2021 – July 2021 Quantitative Researcher Intern
  • Location: JP Morgan Chase & Co. Mumbai, India
  • Duties Included:
    • Built a centralized signal engine platform by designing classes to optimally serve the ML signals live in production
    • Migrated a short-term, mid-price prediction LSTM signal on the built platform, resulting in a 70% reduction in server-side latency (number of CPU’s used)
    • Implemented a stochastic state-space model - Kalman filtering, for Short Interest prediction currently in production
    • Tools used: Python, Keras, Object-Oriented Programming, Reactive Programming, kdb+ database system
• Aug. 2020 – Dec. 2020 Applied Scientist Intern
  • Location: Amazon Machine Learning Hyderabad, India
  • Duties Included:
    • Worked on the transit time estimation in the Last Mile Science team crucial for efficient routing for Amazon deliveries
    • Optimized the transit time using non linear ML models like XGBoost, Random Forest, deep neural networks and introduced meaningful embedding vectors
    • Achieved improvement in Weighted Absolute Percentage Error(WAPE) from 31.96 to 18.78
    • Tools used : Keras, Numpy, Scipy, Scikit-Learn, Pandas, Java, C++, Docker , SQL, Apache Spark
• May 2018 – July 2018 Summer Research Intern
  • Location: Indira Gandhi Centre for Atomic Research(IGCAR) Chennai, India
  • Duties Included:
    • Modeled unique user behaviour like Cognitive Style, interests and the goals using ML algorithms.
    • Proposed and worked on an idea of adding recommendation for a better experience
• July 2018–July 2019 Teaching Assistant
  • Location: BITS Pilani Hyderabad, India, Applied Econometrics, Money Banking and Finance
    • Interacted with students to solve multiple issues with environment setup, incompatible software versions, running software.
    • My duties include preparing and grading programming assignments, holding weekly office hours, answering students questions helping out the professor with other course logistics.
    • Assisting students with technical/programming help pertaining to installations, setup, and issues with R and Python.

Skills

• Programming Languages: Spark, Python, C/C++, Java, R, HTML/CSS
• Deep-Learning Frameworks: PyTorch, Keras, TensorFlow
• Tools: Natural Language Toolkit, Scikit-Learn, Scrapy, AWS, SQL, MS Office, LaTeX, GitHub

Publications

• An Empirical Analysis on the Prediction of Web Service Anti-patterns Using Source Code Metrics and Ensemble Techniques . Published in International Conference on Computational Science and Its Applications (ICCSA), 2021
• WWBP-SQT-lite: Difference Embeddings and Multi-level Models for Moments of Change Identification in Mental Health Forums, runner up at CLPsych 2022 shared task by NAACL
• Economic Monitor: Special Focus Chapter on Digital Disruptions and Economic Impacts by World Bank, 2022

Achievements

• One of the 90 students selected worldwide for Cornell, Maryland, and Max Planck Pre-Doctoral Research School(CMMRS) Summer School that taught about the state-of-the-art research in Computer Science.
• Among the 1% candidates(100 nationwide) to be selected to attend the Google Research India AI Pre Doctoral Summer School, 3 day workshop
• Featured in an interview in the AI Time Journal with Nisha Arya Ahmed Talked about my corporate learnings and future opportunities in AI and advice to someone starting new in the filed.

Projects

• Recommendation System - Information Retrieval
  • Feb 2022 - June 2022
    • Built a deep learning-based recommendation system to predict the rating that a given user gives to an item
    • Applied the Graph Neural Network(GNN) to model the user-user and user-item interactions along with the attention
    • framework to differentiate ratings from the interactions
    • Achived an RMSE of 1.12 and MAE OF 0.85 on Epinions dataset
    • Also implemented CUR and SVD to reduce the dimensionality of the movie lens dataset and implemented user-user, Item-Item collaborative filtering, and latent factor model separately
• Stock Price Prediction using LSTM and deep learning models - Machine Learning, AI
  • Jan 2020 - Mar 2020
    • Forecasted the stock price movement of Amazon by incorporating the sentiments along with the past stock price.
    • GoogleNews API used to capture the latest news description for the day and sentiments extracted using TextBlob
    • Accuracy increased by over 2.5% from 85.2% to 87.5% when the news data was included for prediction using the LSTM model
• Toxic Comment Classification using Transformer based transfer learning
  • Jan 2022 - April 2022
    • Experimented various BERT based architectures and fine tuned the RoBERTa model for the specific task.
    • Currently exploring the techniques to reduce the bias in the false negative class.