Education

  • MSc Data Science for Decision Making, Maastricht University | 2025-2027
    • GPA: 8.3/10 (Cum Laude)
    • Thesis: Physics Informed Machine Learning and Uncertainty Estimation for Crop Yield Prediction, done @ CBS
    • Key courses: Mathematical Optimisation, Computational Statistics, Algorithms for Big Data, Advanced Natural Language Processing, Computer Vision
  • BSc Physics and Mathematics, Bangalore University | 2021-2024
    • CGPA: 9.6/10

Experience

  • Statistics Netherlands (CBS) | August 2026 - February 2027
    • Crop yield estimation from remote sensing inputs; how to make predictions reliable & scalable?
    • Custom CNN based architecture and loss function to improve out-of-domain performance
    • Conformal prediction based framework for uncertainty quantification

  • Reviewer for World Models in Physical AI workshop @ NeurIPS 2026 | September 2026

  • FORGIS | August 2026
    • Action-conditioned world models for time series
    • Developed non-greedy evaluation metric
    • Implemented auto-regressive prediction head
  • Statistics Netherlands (CBS) | February 2026 - August 2026
    • Tree species classification from aerial images; how to work around limited data? (size and modality)
    • Fine-tuned GeoSpatial Foundation models & tried transfer learning from related dataset
    • Derived canopy height maps as scalable proxy for 3D LiDAR information & custom architecture for processing
    • Applied Contrastive Learning as pre-training strategy for image encoder

Hackathons

  • Bunq | Multi-modal AI Hackathon | Demo
    • Guardian AI ; a voice only banking app to save time, reduce errors and increase accessability
    • Responsible for [fraud detection, bill-splitting] features through claude vision and web search tools
  • Veed.io | GenAI and Video Hackathon | $ 2^{nd} $ place | Demo
    • Motif.ai ; create 8 second looping videos for Spotify canvas after analyzing artist’s music
    • Responsible for [text-to-image, image-to-video] models and video editing feature
  • Maastricht University | Supernova buildathon | Demo
    • A note taking platform where students get real-time feedback based on course materials and teachers understand student comprehension
    • Built RAG pipelines for checking notes based on provided context in DataStax
  • Prosus | AI University Games Hackathon | Demo
    • An AI Agent personal assistant that schedules tasks and manages conflicts from natural language voice input and email
    • Built Google Calendar integration tool and Gmail notification scheduler agent

Projects

Chotu

Repo | Demo

  • A multi-agent, markdown based assistant for all things professional

Skills

Programming Languages

  • Python, SQL, R, MATLAB

Frameworks and libraries

  • NumPy, pandas, GeoPandas, matplotlib, scikit-learn, PyTorch, Torchgeo, Terratorch, LangGraph, LlamaIndex

Tools and DevOps

  • Git, Github Actions, Docker

ML Platforms and Frameworks

  • Huggingface (Datasets, Transformers), MLFlow

Certificates

Technical bookshelf