Resume | Hanson Wen

Experience

Current work

Research labs, startup work, and biotech internships shaping the current direction of the work.

Regeneron Pharmaceuticals — Incoming Computational Biology Intern

Summer 2026

Developing predictive models to optimize clinical trial design and patient cohort stratification, leveraging multi-omic data and personalized biology for targeted drug efficacy

UC Berkeley Sky Computing Lab — Student Researcher

Mar 2026–Present

rLLM hive team. Developing Agent Swarms for Biology. [Repo]

Berkeley RDI — Student Researcher

Jan 2026–Present

Helping build AgentHLE, Humanity's Last Exam for computer use agents; benchmarking agent performance across 63 high-GDP domains from 3D modeling to accounting to education

Stealth Biotech — Co-founder

Nov 2025–Present

Building foundational AI models that leverage multi-modal patient input (transcriptomics, histopathology slides, EHR etc.) to help oncologists better predict treatment risks for cancer patients
Model training/inference infrastructure, model architecture design

CytoAurora Biotechnology — Research/ML Intern

Jun–Aug 2025

Contributed to CytoAurora's AI-driven computer vision screening pipeline for circulating tumor cell (CTC) detection from liquid biopsies; worked with nano-biological chip imaging data and assisted in training and evaluating cell classification models for automated histopathology/cytopathology analysis

Education

Academic path

Formal training and the campus environments feeding the biology-computer science overlap.

UC Berkeley — Molecular and Cell Biology + Computer Science

2025–2029

Activities: Project Lead at Computational Biology @ Berkeley, iGEM@Berkeley Computational Team Member (De novo protein binder design, transcriptome foundational models), ACE (Association of Chinese Entrepreneurs), AI Entrepreneurs at Berkeley Incubee SP26

Harrow International School Hong Kong

2012–2025

A-Levels: Math, Further Math, Physics, Biology (A*A*A*A*); SAT: 1570 (790 R/W, 780 M)

Projects

Resume-selected research and builds

The projects that best represent the overlap between technical depth, product judgment, and scientific usefulness.

Personal Genome Analysis

Processed my own genome (131GB WGS data), achieving 88.63% mapping rate; identified 4.99M variants including 471 high-impact mutations across 355 genes; characterized East Asian ancestry (2.08M population-consistent variants); analyzed pharmacogenomic markers (CYP2D6/CYP2C19/CYP2C9) affecting drug metabolism; deployed pipeline on Google Cloud Platform requiring 128GB RAM/1.5TB storage [Repo]

CNV Cancer RNA-seq Analysis

Built reproducible pipeline and web dashboard (CopyKAT, React, FastAPI) to infer and visualize copy number variation patterns from scRNA-seq cancer datasets; separates malignant from normal cells

Enso Atlas

Google MedGemma Impact Challenge Top 10 (Honorable Mention) out of 878 submissions; fully local on-premise pathology AI system for treatment-response prediction; analyzes whole-slide images using TransMIL, Path Foundation, MedSigLIP, and MedGemma for slide-based prediction, evidence visualization/retrieval, and reporting; FastAPI + Next.js + PostgreSQL

Arthritis Biomarker Machine Learning

(DOI: 10.33696/Proteomics.4.015): Identified 79 DEGs across spondyloarthritis and rheumatoid arthritis from spatial transcriptomics of 6 synovial biopsies; 93.3% classification accuracy (LGBM); GO enrichment, KEGG pathway, and SHAP interpretation

De Novo Protein Binder Design & Validation

Built RFdiffusion-based pipeline for generating novel protein structures conditioned on active site motifs; validated designs via structural alignment of catalytic residues using PyMOL and binding affinity scoring with AutoDock Vina; developed as part of iGEM computational team

Awards

Selected honors

A short record of public wins, from technical work to the stranger parts of the story.

Resume and experience.