Mission: Solve as many puzzles as possible.
Computational biologist working the seams between bench biology, structural biophysics, and machine learning at scale. B.S. in Computational Bioengineering @ Northeastern (Dec 2025) and currently pursuing a thesis-track M.S. in Bioengineering (Expected Grad May 2027).
Boston, MA • ashenoycompany@gmail.com • LinkedIn
- scRNA-Flow — end-to-end scRNA-seq pipeline (Scanpy/AnnData): filtering, normalization, doublet detection, integration, clustering, marker gene analysis.
- ATAC-QC — chromatin accessibility QC + peak analysis with TSS enrichment, fragment-size distributions, nucleosome banding diagnostics.
- OmicsQC — sequencing QC across bulk and single-cell modalities; detects low-quality cells, ambient RNA contamination, batch effects.
- OpenADMET-PXR — solo entry in the Hugging Face OpenADMET competition (top 100). 35+ modeling approaches for pEC50 prediction on Pregnane X Receptor, including ChemBERTa fine-tunes, ECFP+LightGBM, and hybrid transformer architectures fusing protein and compound embeddings. Currently exploring activity-cliff-aware pretraining to sensitize representations to small structural changes with large activity shifts.
- ARES + BayesBio — Bayesian inference + ML pipelines for compound prioritization with calibrated uncertainty.
- OrbitSwap (Just Another Studios) — gravity-puzzle game shipped on Android. Kotlin, original mechanic design.
| Where | When | What |
|---|---|---|
| UCB Biosciences (Computational Data Science Co-op) | 2025 | Multimodal drug-discovery data; UMAP + SHAP-driven artifact detection; ML compound prioritization. |
| COMBINE Lab, Northeastern (Undergrad Researcher, Dr. Mona Minkara) | 2023–2025 | Reproducible HPC pipelines on SLURM; molecular modeling; 15+ technical talks; 2 ongoing manuscripts. |
| Arbor Biotechnologies (Gene Therapy Co-op) | Jan–Jun 2024 | DOE-driven AAV production optimization; >250% productivity improvement. |
My emerging research thesis: knowledge graphs are the right substrate for integrating multi-omics. Clinical, molecular, phenomic, metabolomic, pathway, and simulation data each carry the experimental context that produced them. AI agents traversing those graphs can surface connections that no single human can hold in their head — which matters most in domains like immunology, where heterogeneity, tissue context, and patient variability make the problem combinatorial.
If you're working on anything in that direction, I'd love to hear from you.
- AAAS 2025 (American Association for the Advancement of Science) — presenter
- RISE Northeastern — 2024, 2025
- MBN Biophysics Conference — 2024
- Source Fair, Northeastern — presenter
Python (pandas, numpy, scipy, scikit-learn, Scanpy, AnnData, PyTorch) • R • SQL • Bash/SLURM • C++ • Kotlin • Docker • Conda • Jupyter • Cursor • Claude Code • Schrödinger Maestro/Glide/Prime • RDKit • Benchling
"The interesting questions in biology live in the gaps between disciplines."