Skip to Content

Hi, my name is

Sugam Budhraja.

I'm an AI trainer (with a PhD).

I coach machines to hit their targets,
and make smarter decisions.

About Me

Introduction


I'm an AI researcher and engineer, originally from India and now based in New Zealand.

With a PhD in AI and experience building across startups and MNCs, I've spent the past few years turning ML research into real-time, scalable, and explainable AI products.

Outside of work, I sketch, hike mountains, and lose to chess engines in increasingly creative ways.

Headshot

My Work Experience

  • April 2023 - Present

    Head of Data Science @ Sahha

    • Led the strategy and development of Sahha’s core AI products—health scores, biomarkers, and archetypes—shaping roadmap, research, and go-to-market direction.
    • Built real-time, scalable pipelines and explainable AI models for health prediction from passive smartphone data.
    • AWS
    • DBT
    • Databricks
    • MLFlow
    • Postgres
    Auckland, NZ
    Head of Data Science
  • June 2021 - February 2022

    Machine Learning Engineer @ EyeInc

    • Led the development of a mobile-based eye-tracking system, fine-tuning Google’s GazeNet to achieve ~1.5° accuracy—approaching clinical-grade precision at a fraction of the cost.
    • Built and deployed a real-time gaze data pipeline and edge-optimized models, enabling scalable attention and behavior analytics from smartphones.
    • PyTorch
    • CoreML
    • Swift
    • Xcode
    • Shopify
    Auckland, NZ
  • May - July 2019

    Full Stack Intern @ Intuit

    • Built a graph database-based backend system on AWS Neptune with REST APIs for querying service-to-service interactions and insights.
    • Designed a web app to visualize graph structures, delivering a full-stack solution for developer usability and analytics.
    • Spring Boot
    • React.js
    • Bootstrap
    • Gremlin
    Bengaluru, IN
  • May - July 2018

    Data Science Intern @ Reliance Jio

    • Built a facial recognition system with spoof detection, combining models like FaceNet, Dlib, and DeepFace for secure access control.
    • Optimized face matching to reduce inference time from 500ms to 6ms, enabling deployment in high-throughput environments.
    • Python
    • C++
    • Keras
    • OpenCV
    • Dlib
    Mumbai, IN

Some Things I’ve Built

  • Folder

    NeuroGeMS

    Interactive GUI software for multimodal biomedical research. Combines multimodal model design and MLflow experiment tracking in a seamless research workflow.

    • Flask
    • React.js
    • Material UI
    • MLflow
  • Folder

    NeuWave

    GUI Software for EEG brain data analysis, classification and interpretation with time-frequency and connectivity feature extraction, and ML and DL models.

    • MNE
    • PyTorch
    • Numpy
    • Matplotlib
  • Folder

    NeuCube-Py

    Python toolkit for NeuCube framework. Supports spatio-temporal brain data modeling using spiking neural networks.

    • PyTorch
    • SnnTorch
    • Numpy
  • Folder

    Information Retrieval

    Backend of a software that retrieves information from the web or a custom corpora (large collection of documents). Uses term weighting (tf-idf) and clustering to rank the documents.

    • Python
    • NLTK
    • Beautiful Soup
  • One Shot Learning

    Implemented a siamese network (based on paper) for classification of images from just one training sample. Obtained 91% accuracy on the Omniglot dataset. 1st place in Kaggle competition.

    • Python
    • Keras
    • NumPy
  • Folder

    Intelligent Medicine Box

    A software that uses weight sensors on a Raspberry Pi-enabled medicine box to detect anomalies in pill consumption. Supports notifications to the patient and doctor through companion android app.

    • Python
    • Firebase
    • Android

What’s Next?

Get In Touch

I love connecting with people and I am always open to new opportunities. Whether you have a topic to discuss, a question to ask or just want to say hi, drop me an email and I'll get back to you!