Computational Biology • Bioinformatics • AI Research

Building intelligent systems for
biology and medicine.

Bioinformatics researcher and AI engineer focused on graph machine learning, computational biology, explainable AI, and LLM evaluation. Author of research spanning transcriptomics, molecular toxicity prediction, healthcare AI, and generative AI systems. Based in India.

Get in touch → LinkedIn GitHub Scholar
Portrait of Gurojaspreet Kaur
Gurojaspreet Kaur · AI Research
6
Research outputs
~500
NLQs reviewed
4
First-author works
2
Active bioinformatics papers
01 —

Experience

Software Engineer Intern — GenAI Evaluation & AI Reliability
IQVIA
  • Designed and maintained evaluation datasets and Ground Truth pipelines for enterprise LLM workflows; reviewed ~500 natural language queries to improve semantic accuracy and downstream model reliability.
  • Built automated benchmarking and prompt-testing loops in Python to evaluate latency, structured output correctness, response consistency, and semantic reliability across Mistral-based models.
  • Conducted systematic failure analysis and prompt-level optimization to improve production GenAI workflow stability.
  • Developed internal investigation tooling and reproducible documentation to accelerate diagnostic speed across evaluation workflows.
Jun 2025 – Dec 2025
India
Independent Technical Research Consultant
Freelance
  • Authored technical research content across machine learning, artificial intelligence, healthcare AI, and cloud technologies.
  • Supported academic publications, technical documentation, and research communication projects.
  • Translated complex engineering and research concepts into accessible material for multidisciplinary audiences.
  • Produced long-form technical content covering deep learning, NLP, generative AI, and data science applications.
Aug 2024 – Present
Remote
02 —

Selected Projects

Bioinformatics
Comparative Evaluation of Deep Tabular and Generative Language Models for Breast Cancer Subtyping
Compared TabNet and LLM-based approaches on high-dimensional transcriptomic data; reduced 54,675 genes to 10,000 features and achieved 96.77% accuracy with attention-based deep tabular learning.
ICDAM 2026
GenAI
LLM Evaluation & GenAI Workflow Benchmarking
Built evaluation workflows for enterprise LLM systems, benchmarking semantic reliability, response consistency, structured output correctness, and latency across production-facing GenAI use cases.
IQVIA
GNN / XAI
Evaluating the Role of Attention-Based Message Passing in Explainable Molecular Toxicity Prediction
Submitted independent research comparing isotropic GCN and attention-based GAT message passing for explainable molecular toxicity prediction, using GNNExplainer to generate bond-level toxicophore attribution masks on the Tox21 dataset.
Under Review
Comp Bio
PathXComm: A Pathway-Aware Explainable Graph Neural Network Framework for Cell–Cell Communication Inference in the Tumor Microenvironment
Ongoing research integrating pathway-aware biological priors, explainable graph neural networks, and tumor microenvironment modeling for cell–cell communication inference.
Ongoing
Finance AI
Real-Time Adaptive Multi-Modal Stock Prediction
Developed a temporal graph attention framework combining financial signals, sentiment-aware market modeling, volatility-sensitive reweighting, and event-triggered attention mechanisms.
IEEE CAI
NLP
Hate Speech Detection Using Ensemble Learning
Built an ensemble machine learning pipeline using SVM, Naive Bayes, and Decision Trees for online hate speech detection, achieving 89.53% accuracy on benchmark datasets.
AIP
03 —

Papers & Publications

Sole Author
Evaluating the Role of Attention-Based Message Passing in Explainable Molecular Toxicity Prediction
Under review at Computational Biology and Chemistry — Elsevier, Q2 in Bioinformatics Domain
Under Review
Sole Author
PathXComm: A Pathway-Aware Explainable Graph Neural Network Framework for Cell–Cell Communication Inference in the Tumor Microenvironment
Independent Research — In Development, 2026
Ongoing
First Author
Comparative Evaluation of Deep Tabular and Generative Language Models for High-Dimensional Transcriptomic Subtyping in Breast Cancer
ICDAM 2026 — Springer LNNS
Accepted
First Author
Review of Deep Reinforcement Learning and Artificial Neural Networks in Healthcare Systems
Elsevier Book Chapter, 2025
Published
First Author
Hate Speech Detection Using Machine Learning: An Ensemble Technique
American Institute of Physics (AIP), 2025
Published
Co-Author
Real-Time Adaptive Multi-Modal Stock Prediction with Temporal Graph Attention and Dynamic Interaction Networks
IEEE Conference on Artificial Intelligence (CAI), 2025
Published
04 —

Technical Skills

Languages
Python SQL C++ Java
Cloud & Infra
GCP Azure Docker Kubernetes MLflow Linux
ML Frameworks
PyTorch TensorFlow Keras HuggingFace Scikit-learn XGBoost
AI/ML Domains
LLM Eval RAG GenAI NLP Computer Vision GNN XAI
Specialized Libs
PyG RDKit TabNet LoRA/PEFT GNNExplainer
Dev Tools
Git Jupyter VS Code MLflow
05 —

Research Interests

Computational Biology
Bioinformatics Cancer Genomics Transcriptomics Cell–Cell Communication Precision Oncology
Molecular AI
Drug Discovery Molecular Toxicity Toxicophore Localization RDKit
Modeling
Graph Neural Networks Explainable AI Tabular Deep Learning Attention Models
GenAI
LLM Evaluation RAG Prompt Testing AI Reliability