Lagunita Theme


Welcome to my site - a collection of my Stanford experiences and other personally significant things.

Technical Experiences

  • Quantitative Research Intern - IMC Trading
    06/23 - 09/23

  • SDE Intern - Meta [Advertiser Products – Optimization Incubation]
    06/22 - 09/22

    • Developed multitask learning models for ad targeting and optimization for org-wide impact in Caffe2 and DPER3 framework.
    • Tested new architecture through extensive experimentation to prove efficacy both in isolation and as a sub-module of a larger model.
    • Updated existing pipelines to expand the model's use cases across two datasets.

  • Undergraduate Researcher - Stanford Institute of Economic and Policy Research
    03/22 - present

    • Collected data concerning the FCC 904 RDOF auction, which subsidized providing ~5M homes and businesses with high-speed internet.
    • Formulated research plan to analyze bidder behavior and FCC cost models as a function of intrinsic bidding lot factors such as geospatial distribution and demographic factors.
    • Processed data to analyze the effect of neighboring census block groups on auction results.
    • Developed methodology to measure bid to footprint distance across all bids.



  • AI/ML Intern -
    10/21 - 06/22

    • Modeled query data with synthetic language. Developed API interfacing with CosmosDB versioned dataset for dataset maintenance.
    • Experimented on multiheaded named entity extraction NLP model through methods such as cross dataset transfer learning.

  • SDE Intern - Amazon [Music]
    6/21 - 9/21

    • Developed an automated pipeline for delivering music publisher streaming reports incorporating additional required catalog data.
    • Leveraged Spark Scala for optimized distributed computation due to expanded data throughout.
    • Performed thorough testing using Airflow to ensure success for a variety of business-critical configurations.

  • Undergraduate Researcher - Stanford AI Lab [ML Group]
    6/20 - 6/21

    • Led project to develop a novel semi-supervised contrastive learning methodology leveraging sample metadata for pair synthesis.
    • Applied algorithm to automatic lung and heart sound diagnosis from electronic stethoscope audio data.
    • Synthesized experimental regimes, developed scripts and code, and compiled experimental results for publication.

  • Undergraduate Researcher - Stanford AI Lab [Dror Lab]
    6/20 - 5/23

    • Researched the use of self-attention in graph neural networks with applications in protein-ligand pose prediction.
    • Developed PyTorch package to open-source reproducible code for minimally equivariant graph neural networks.
    • Developed Syntactically Aware Emeddings using EGNNs to embed syntactic information in foundation model embeddings.


  • Back End Developer
    5/20 - 12/21

    • Built a startup dedicated to building communities for parents of children with mental illnesses.
    • Constructed content delivery back-end architecture and Rest API utilizing Django framework.
    • Co-led engineering team-wide reviews for technical product alignment between front-end and back-end development.

  • Research Trainee - Psychiatry Neuroimaging Lab, Harvard Medical School
    6/19 - 8/19

    • Completed an AI research project under the direction of Dr. Suheyla Cetin-Karayumak as part of the Research Science Institute (RSI), hosted at the Massachusetts Institute of Technology (MIT) by the Center for Excellence in Education (CEE).
    • Built a deep learning model, Flexible Medical Image Super-Resolution (FMISR), to enhance MRI quality using Python/Tensorflow and trained/tested the model with real-world conditions to measure clinical viability.
    • FMISR uses skip-connect layers and a residual block architecture to improve model performance.


  • Research Intern - Henry Ford Health System Bone & Joint Center
    6/18 - 8/18

    • Algorithmically measured bone properties from Digital Tomosynthesis (DTS) scans of in-vitro and in-vivo vertebras using Digital Volume Correlation analysis under the direction of Dr. Yener Yeni.
    • The method is promising as a way to monitor bone quality for osteoporosis diagnosis and treatment.
    • Co-author of two abstracts accepted to the Orthopedic Research Society.

  • Independent Research
    6/17 - 3/18

    • Developed a novel context-aware convolutional neural network model (C-SRCNN) to enhance the quality of CT scans using Python/Tensorflow. Explored multiple methods of incorporating context for an expanded effective field, boosting performance compared to prior models.
    • Validated performance on general and medical images.



2022 - 2023

  • Spring
    • STATS 361: Causal Inference
    • ECON 291: Social and Economic Networks
    • ECON 102C: Advanced Econometrics
    • Math 115: Real Variables
  • Winter
    • CS 224W: Machine Learning with Graphs
    • CS 238: Decision Making under Uncertainty
    • MS&E 232H: Introduction to Game Theory (Accelerated)
    • MS&E 211: Introduction to Optimization
    • STATS 315A: Modern Applied Statistics: Learning
  • Fall
    • CLASSICS 37: Great Books, Big Ideas from Ancient Greece and Rome
    • ECON 157: Imperfect Competition
    • MS&E 193: Technology and National Security
    • MS&E 245A: Investment Science
    • STATS 200: Introduction to Statistical Inference

2021 - 2022

  • Spring
    • CS 231N: Deep Learning for Computer Vision
    • ECON 165: International Finance
    • ECON 108: Data Science for Business and Economic Decisions
    • ECON 137: Decision Modeling and Information
    • MATH 113: Linear Algebra and Matrix Theory
  • Winter
    • CS 110: Principles of Computer Systems
    • ECON 143: Finance, Corporations, and Society
    • ECON 136: Market Design
    • CS 224N: Natural Language Processing
    • STATS 217: Introduction to Stochastic Processes I
  • Fall
    • CS 109: Introduction to Probability for Computer Scientists
    • STATS 116: Theory of Probability
    • ENGR 40M: An Intro to Making: What is EE
    • CEE 63: Weather and Storms
    • PWR 2HK: Writing & Rhetoric 2: Think Global: The Rhetoric of Global Citizenship

2020 - 2021

  • Summer
    • ECON 102B: Applied Econometrics
  • Spring
    • CS 221: Artificial Intelligence: Principles and Techniques
    • CS 229: Machine Learning
    • ECON 51: Economic Analysis II
    • ECON 52: Economic Analysis III
    • THINK 48: Reading the Body: How Medicine and Culture Define the Self
  • Winter
    • CS 107: Computer Organization and Systems
    • CS 230: Deep Learning
    • ECON 102A: Introduction to Statistical Methods (Postcalculus) for Social Scientists
    • HUMBIO 174: Intro to Bioethics
    • MUSIC 220B: Compositional Algorithms, Psychoacoustics, and Computational Music
    • DANCE 1: Contemporary Modern I: Liquid Flow
    • DANCE 58: Hip Hop I: Introduction to Hip Hop
  • Fall
    • CS 106B: Programming Abstractions
    • MATH 53: Ordinary Differential Equations with Linear Algebra
    • ECON 50: Economic Analysis I
    • CS 522: Seminar in Artificial Intelligence in Healthcare
    • PWR 1HF: Writing & Rhetoric 1: From Ghost Bikes to the Googleplex: Digital Rhetoric and Social Action


  • BallotProof
    Built computer vision methods for BallotProof, a ballot checker that performs all computations client-side for maximum security. BallotProof received a top8 award at HackGT 7. Test files can be found on our GitHub Repo.



  • COVID-19 Transmission Simulator
    Created a JavaScript simulator to illustrate the impact of respiratory protection and quarantine policies on transmission initiated by medical professionals. The simulator leverages agent-based modeling to emulate real-world transmission possibilities.


  • Tech2Serve
    Empowering the underserved through technology. Projects include smart devices for visually impaired students, India's first E-Braille Library, laser technology for artificial limb production, educational aids for students with intellectual disabilities, and financial support for unemployed visually impaired individuals. Developing technology clubs in Detroit schools and a wearable ECG-CPR device to monitor and save at-risk hearts.


Teaching/Other Experiences

  • InspiritAI
    12/20 - Present
    Instructor for Inspirit AI, delivering AI education and practical project experience to high schoolers. Also, creating new projects as a curriculum developer.

  • Crimson Counseling
    12/19 - Present
    Founded business to consult and mentor early high schoolers to develop strong, balanced habits for high school success.

  • The Tutoring Center
    6/18 - 3/19
    Tutored math and English individually and in small groups to students pre-K to 12th grade. Was responsible for housekeeping duties.

  • Private Tutoring
    12/17 - Present
    Tutored competition math and school math to students of all levels in group and individual sessions.

Accolades & Publications


  • Contrastive Learning Leveraging Patient Metadata for Sample-efficient Lung and Heart Sound Representations
    Cell: Patterns
    Pratham Soni ; Siyu Shi ; Pranav Sriram ; Andrew Ng ; Pranav Rajpurkar


  • Equivariant Graph Neural Networks for 3D Macromolecular Structure
    International Conference on Machine Learning CompBio Workshop 2021 (Spotlight)
    Bowen Jing ; Stephan Eismann ; Pratham Soni ; Ron O. Dror


  • Tomosynthesis-Based Digital Volume Correlation Properties Predict Vertebral Strength Independently from Bone Mineral Density
    Conference Proceedings of the Orthopedic Research Society 2020
    Joshua P. Drost ; Daniel Oravec ; Pratham Soni ; Roger Zauel ; Michael J. Flynn ; Yener Yeni


  • Clinical measurement of vertebral stiffness and displacements using tomosynthesis based digital volume correlation
    Conference Proceedings of the Orthopedic Research Society 2019
    Daniel Oravec ; Pratham Soni ; Roger Zauel ; Sudhaker Rao ; Michael J. Flynn ; Yener N. Yeni



  • HackGT (Georgia Tech) - Top 8 projects overall
  • Summer Cohort - AIHC