terminal -- 80x24

Last login: 9/18/2025 11:36:06 AM

ananya@yale-university:~$ whoami

_

Experience

Research & Entrepreneurship in AI x Bio

LinkedIn
Digital effect
Generative AI Scientist
University of Pennsylvania - Pranam Chatterjee Lab
May 2025 - Present

Leading research in generative AI applications for computational biology and drug discovery.

Generative AI
Deep Learning
Computational Biology
Python
Machine Learning Researcher
IBM Thomas J Watson Research Center
Dec 2024 – Present

Fine-tuning multimodal foundation models with contrastive learning approaches to enhance molecular tokenization and representations.

Multimodal AI
Foundation Models
Contrastive Learning
Transformers
Geometric Deep Learning Researcher
Yale University - Smita Krishnaswamy Lab
Sept 2024 – Present

Develop geometric deep learning models for dynamic regulatory interaction graphs from time series data with perturbations.

Paper
Geometric Deep Learning
Graph Neural Networks
Multi-Agent RL
Time Series
Deep Learning and Interpretability Researcher
GSK (GlaxoSmithKline)
May 2024 – Aug 2024

Created deep learning based thermophilicity predictor with protein language models for directed evolution and improved biocatalysis reactions.

Protein Language Models
Explainable AI
Directed Evolution
Biocatalysis
Machine Learning Developer & Entrepreneurial Lead
National Science Foundation
Jan 2023 – Present

Entrepreneurial lead and acting CEO of lab-based startup using machine learning and meta-modeling for drug discovery.

Entrepreneurship
Grant Writing
Business Development
Meta-modeling
Software Developer for Computational Drug Discovery
Yale University - David Spiegel Lab
Jan 2023 – Present

Python and bash algorithms to automate iterative molecular docking and enable virtual high throughput hit screening with Monte Carlo Tree Search.

Lab Website
Python
Bash
Molecular Docking
Monte Carlo Methods
Computational & Structural Biochemistry Researcher
National Institutes of Health (NIAID)
May 2023 - July 2023

Harnessed computational modeling tools and wet-lab techniques to design, develop, and evaluate nanoparticle vaccines for malaria and COVID-19.

Structural Biology
Vaccine Design
Protein Purification
ELISA

Projects

Fun stuff I've built!

GitHub
Digital effect
Group Relative Linear Markov Decision Processes
Inspired by group relative policy-based approaches, developed two types of approaches to sampling in LMDPs with aggregated features and group-relative sampling. Saw faster learning across common benchmarks such as GridLock.
Paper
S&DS 685: Theory of Reinforcement Learning
Grade: A
Reinforcement Learning
Markov Decision Processes
Policy Optimization
Python
Explainable AI for Protein Function Prediction
Directed team in evaluating AI based protein functional prediction models. Created GradCAM, Excitation Backpropagation, and PG Explainer models with random adversarial perturbation to assess most deterministic structural motifs.
Poster
CPSC 574: Trustworthy Deep Learning
Grade: A
Accepted to Interpretable Systems for Artificial Intelligence Transparency
Presented at Harvard/MIT Machine Learning for Drug Discovery Symposium
Explainable AI
Graph Neural Networks
Protein Function
GradCAM
Deep Learning
Dynamic Curvature Optimization for Hyperbolic GCNs
Developed novel optimization methods for learnable manifold curvature in Hyperbolic Graph Convolutional Networks, showing 9% improvement over state of the art methods.
PaperPoster
CPSC 583: Deep Learning on Graph Structured Data
Grade: A
Presented at International Conference on AI and Mechatronics
Hyperbolic Geometry
Graph Neural Networks
Manifold Learning
Optimization
Circuit Tracing with Transcoders for Protein Mechanistic Interpretability
A comparison to Sparse Autoencoders on Evolutionary Scale for understanding protein mechanistic interpretability through circuit tracing methodologies.
Visualizer
Mechanistic Interpretability
Transcoders
Sparse Autoencoders
Protein Analysis
Defensive Stochastic Activation Pruning on Deep Learning Architectures
Tested and demonstrated efficacy of various stochastic activation pruning approaches on ResNet-18 to defend against Projected Gradient Descent and Fast Gradient Sign Method attacks in image classification.
Adversarial Defense
Neural Network Pruning
ResNet
Computer Vision
Adversarial AI + Simulated Network Threat Detections
Custom Go/Powershell C2 tooling, Designing Sigma & Suricata detection rules, and automated lab deployment for AI attacks.
AI AttacksAdversary Sim
2024
Adversarial AI
Network Security
Threat Detection
Go
PowerShell
C2 Framework
Digital effect

Publications

Always a work in progress.

STAGED: A Multi-Agent Neural Network for Learning Cellular Interaction Dynamics
arXiv preprint, 2024
2024

Authors: Joao F. Rocha, Ke Xu, Xingzhi Sun, Ananya Krishna, Dhananjay Bhaskar, Blanche Mongeon, Morgan Craig, Mark Gerstein, Smita Krishnaswamy

DOI: arXiv:2507.11660

Read Paper
DeepFRI Demystified: Interpretability vs. Accuracy in AI Protein Function Prediction
Interpretable Systems for AI Transparency Workshop at CVPL International Conference on Image Analysis and Processing, 2025
Sapienza University, Rome, September 2025
2025

Authors: Ananya Krishna, Valentina Simon, Arjan Kohli

Various manuscripts in preparation...

About Me

Feel free to reach out!

Ananya Krishna

Ananya Krishna

AI Researcher

Passionate about solving complex scientific and security-related problems through the lens of artificial intelligence and computational methods.

Research Interests

  • Interpretability, Explainability, and Adversarial AI/Defenses
  • Generative Biological Design with AI
  • Optimization and Computation
  • AI-driven Drug Discovery
  • Biosecurity and Engineered Pathogens

Personal Interests

  • Hiking and Weight Lifting
  • Science Fiction Literature
  • Classical Viola, Opera, Music Composition (10+ years)
  • Open Source Software Development
  • Calligraphy, Creative Writing, and Wax Seal Making

The inherent lyricism of science demands a rhythmic precision, but one that must be paired with expressive creativity.