Skip to content

UC Berkeley · Chem Bio × CS

Andrew Boldi

I solve the hardest problems in science with computation — teaching machines the language of molecules, proteins, and matter.

Selected work Who I am
  • medicinal chemistry
  • drug discovery
  • organic synthesis
  • structural biology
  • deep reinforcement learning
  • molecular machine learning
  • biochemistry
  • explainable AI
  • foundation models
  • computational chemistry
  • electronics & making
  • total synthesis
  • medicinal chemistry
  • drug discovery
  • organic synthesis
  • structural biology
  • deep reinforcement learning
  • molecular machine learning
  • biochemistry
  • explainable AI
  • foundation models
  • computational chemistry
  • electronics & making
  • total synthesis
// status

Currently an AI intern at Soley Therapeutics doing molecular machine learning for drug discovery — and finishing a B.S. in Chemical Biology and a B.A. in Computer Science at UC Berkeley.

  • AI Intern Soley Therapeutics
  • UC Berkeley Chem Bio · CS
  • Class of June 2027

By the numbers

0 fastest 3×3 solve
0 AMC Trainer users
0 countries reached
0 students taught
0 piano competition wins
0 ADMET datasets benchmarked
0 consecutive synthesis steps
0 Street View images mined

What I work on

Six fields, one question:
how does matter compute?

01

Molecular machine learning

Foundation models, finetuning, and explainability for molecular property prediction — turning chemistry into something a network can reason about.

02

AI-driven drug discovery

Generative chemistry and lead optimization: searching latent chemical space for molecules that are novel, synthesizable, and actually do something.

03

Total synthesis

Multi-step organic synthesis at the bench (Stoltz Lab) and deep RL for planning routes — building complex molecules atom by atom.

04

Structural biology

Protein structure and the machine-learning revolution around it — folding, function, and the geometry that connects them.

05

Physics & computation

Statistical mechanics, quantum chemistry (DFT in ORCA), and the optimization landscapes that ML and the physical world quietly share.

06

Engineering & making

Electronics, M68k assembly, 3D printing, robotics automation, and shipping software people actually use. A polymath’s toolbox.

Selected work

Things I’ve built
and figured out.

All work →
01 Trennen Predicting the direction of optical rotation for chiral molecules with a T-Net architecture — distilling 8M+ PubChem molecules to ~27K optically active ones, with DFT polarizability tensors (B3LYP/def2-SVP). Machine LearningChiralityDFTRDKit 02 AMC Trainer An interactive math-competition trainer built with my brother Ethan — adaptive AMC practice now used by 20K+ people across 50+ countries. WebEducationCloudflare 03 Pathfinder A scenic-routing algorithm (modified A*) over San Mateo County — semantic segmentation of 250K+ Street View images on TPUs, with earth-curvature-aware heuristics and TIGER/Line road graphs. Deep LearningComputer VisionGeospatial

Writing

From the notebook

Unroll the essays →

Bookshelf

What I’m reading

A living shelf of the books shaping how I think — biology, ML, physics, chemistry, and the occasional detour. Each with a few honest words.

Open the shelf →

What’s next

Let’s build something
that matters.

Who I am The long story Background, the throughline, beyond the lab. Read about me What I build Selected work Research, tools, and things I figured out. See the work
or just say hello andrewjboldi@gmail.com