Santanil Jana
Mathematics PhD · Quantum Computing Researcher · Machine Learning Scientist

Santanil Jana

I work at the intersection of quantum computing, algebraic topology, graph-based inference, and machine learning. My research focuses on turning deep mathematical structure into scalable computational methods for high-dimensional scientific problems.

View Research AI + ML Projects
Vancouver, BC santanil.jana@gmail.com LinkedIn GitHub

Research

My research combines pure mathematics, quantum information, and computational methods, with a focus on structure, classification, and scalable inference.

Quantum Computing

As a postdoctoral researcher, I study quantum nonlocality, contextuality, and magic state distillation through algebraic, graph-theoretic, and computational techniques.

  • Developed graph-based frameworks for modeling quantum nonlocality as constraint satisfaction problems.
  • Classified three-qubit nonlocality paradoxes beyond the GHZ scenario, identifying new families of paradoxes and their algebraic structure.
  • Developed invariant-theoretic methods for magic state distillation, scaling analysis to codes up to 71 qubits / 41 qutrits.

Algebraic Topology

My PhD research focused on algebraic topology, especially cohomological methods related to unordered flag manifolds and symmetric group actions. This work develops structural tools for understanding cohomology rings of unordered flag manifolds and related algebraic invariants. For more on this research, see my PhD Thesis.

  • Cohomology of complete unordered flag manifolds.
  • Mod-2 cohomology rings of alternating subgroups of Coxeter groups of Type B.
  • Connections between group cohomology, topology, and combinatorial algebra.

AI + Machine Learning

Applied machine learning projects spanning LLM evaluation, deep learning compression, NLP, predictive modeling, and data-driven infrastructure analysis.

LLM Hallucination Detector

Erdős Institute Data Science Boot Camp · Spring 2026

    We built a hallucination detection framework for Transformer-based generative models using Python, PyTorch, and Scikit-Learn. The project combined white-box and black-box signals by extracting spectral features from attention-head Laplacians and geometric features from sampled response clouds. Across 2,500 queries and five benchmark datasets, the framework achieved 0.91 AUROC and served as a lightweight pre-validation layer for identifying risky LLM outputs before downstream verification.
Transformers PyTorch LLMs Model Evaluation Scikit-Learn

Tensorized CNNs for Amazon Deforestation Analysis

Multiverse Computing via PIMS M2PI · Summer 2023

    We optimized a deep learning pipeline for Amazon deforestation analysis by replacing ResNet50 with Tensorized CNN architectures, reducing the model size by approximately 90%. This approach achieved roughly a 10x reduction in FLOPs while maintaining comparable predictive performance on high-dimensional image data. The project involved analyzing around 5,000 images using TensorFlow/Keras and reduced training time by approximately 40%.
Deep Learning Tensorized CNNs Computer Vision TensorFlow Keras

NLP and Predictive Analytics at IOTO International

Data Scientist Internship · 2022

    I developed NLP pipelines over 50,000+ unstructured text records using Scikit-Learn and Pandas, improving classification accuracy by approximately 25% while reducing preprocessing time by approximately 40%. I also queried over 1M structured rows using SQL and Python, building predictive models that improved key business metrics by approximately 15%–20%.
NLP SQL Pandas Predictive Modeling Data Science

EV Charger Capacity and Grid Infrastructure Optimization

Awesense via PIMS M2PI · Summer 2022

    We processed and curated large-scale grid data using SQL and Python, reducing data inconsistencies by approximately 20% and engineering predictive features for capacity modeling and infrastructure planning. I also developed a real-time data pipeline interfacing with the Awesense API to simulate 100+ grid stress scenarios.
API Integration Simulation SQL Scikit-Learn Infrastructure Analytics

Skills

Technical strengths across machine learning, scientific computing, and mathematical research.

Programming & Tools

Python, PyTorch, NumPy, Pandas, SciPy, Scikit-Learn, TensorFlow/Keras, SQL, Git, R, Mathematica, Magma.

Machine Learning

Transformers, LLMs, CNNs, representation learning, regularized regression, classification, clustering, statistical learning, model evaluation.

Scientific Computing

Graph algorithms, constraint satisfaction, uncertainty-aware validation, high-dimensional analysis, mathematical modeling, experimental design.

Research Areas

Quantum computing, quantum nonlocality, magic state distillation, algebraic topology, group cohomology, graph theory.

Publications

  1. 1 M. Zurel, SJ, N. de Silva. High-threshold magic state distillation with quantum quadratic residue codes, Preprint, 2026. ePrint: arXiv:2603.18560
  2. 2 N. de Silva, SJ, M. Yin. Three-qubit nonlocality paradoxes: beyond GHZ, Preprint, 2026.
  3. 3 N. de Silva, SJ, M. Yin. A Classification Program for Nonlocality Paradoxes of Three Qubits, Proceedings of Quantum Physics and Logic, 2025. ePrint: arXiv:2508.14673
  4. 4 L. Guerra, SJ. Cohomology of Complete Unordered Flag Manifolds, Transactions of the American Mathematical Society, 2025. ePrint: arXiv:2309.00429
  5. 5 L. Guerra, SJ, A. Maiti. Mod-2 Cohomology of Unordered Flag Varieties in Lower Orders and Auerbach Bases, Topology and its Applications, 2025. ePrint: arXiv:2304.12990
  6. 6 L. Guerra, SJ. The mod 2 cohomology rings of the alternating subgroups of the Coxeter groups of Type B, Homology, Homotopy and Applications, 2026. ePrint: arXiv:2312.12346
  7. 7 SJ. On the Cohomology of the Total Space of Classifying Space for Commutativity in U(3), 2023. ePrint: arXiv:2106.09195

Experience

Quantum Computing Research, Data Science, and Scientific Computing Experience.

Simon Fraser University

Postdoctoral Research Scientist · 2023–Present · Burnaby, BC

Working as a postdoctoral researcher under Dr. Nadish de Silva, Canada Research Chair in the Mathematics of Quantum Computation. Conduct research in quantum computing, with a focus on nonlocality, contextuality, and magic state distillation. Develop computational frameworks for classifying structured quantum phenomena and evaluating high-dimensional scientific models.

IOTO International

Data Scientist Intern · 2022 · Vancouver, BC

Built NLP and predictive analytics pipelines using Python, SQL, Pandas, and Scikit-Learn. Delivered data-driven insights through classification models, automated reporting, and executive-facing analysis.

Contact