A first automation
A robotic-process hack, then a website for a public-vote contest. Lesson one: scrappy execution beats perfect plans.
Rachel Hu.
Helping everyone turn curiosity into decisions that matter.
Now building · YC S23 Energent.ai ↗ AI that audits AI
Applied scientist
Teacher
Speaker
Founder
CEO
TeacherSpeakerFounderCEO
01 — A working principle
“Fall in love with the problem, not the solution.”
At AWS, Rachel watched machine-learning teams lose 60–80% of their time to messy, unstructured data. Usability, not models alone, is what turns AI into impact. Everything since has been about closing that gap.
♥ the problem, every wall of it ✕ no shortcut drawn
02 — The path
Age 12
Mathematics, with honours
University of Waterloo. President's Scholarship and the Mathematics Faculty Dean's Honours List.
Risk, priced
Society of Actuaries exams in Probability and Financial Mathematics, later GARP's FRM Level I.
Into deep learning
Machine Learning, the Deep Learning Specialization and Google Cloud data engineering.
AWS AI & D2L.ai
Applied scientist at AWS. Builds LLMs and helps lead Dive into Deep Learning, used at 500+ universities.
On stage
NVIDIA GTC, a KDD tutorial, the Toronto ML Summit and computer vision at Machine Learning University.
CambioML, YC S23
Leaves AWS to co-found CambioML with Kimi Kong, in Y Combinator's Summer 2023 batch.
Energent.ai
An AI analyst with a built-in auditor that checks its own work against the source.
Hub71, Abu Dhabi
A second home base, and Demystifying Hub71 for founders weighing the move.
03 — Teaching
Dive into
Deep Learning.
An interactive, open-source textbook with runnable code, maths and discussion. Rachel helped lead it at AWS and taught it on stage, from NVIDIA GTC to Amazon's Machine Learning University.
0+
universities teach with it
Read it at d2l.ai ↗
04 — On stage
Explaining hard things out loud.
NVIDIA GTC, KDD, the Toronto ML Summit and Amazon's Machine Learning University. Hover a talk to see it, click to open.
05 — Writing
Notes from the founder's desk.
06 — Now building
AI that audits AI.
- 01Bring the mess
Spreadsheets, PDFs, decks, engineering drawings.
- 02The analyst works
An AI analyst does the data work and drafts the answer.
- 03The auditor checks
A second agent with fresh context re-derives figures and traces each number to its source.
- 04Pass, with evidence
A verdict and an evidence trail, before anything reaches a decision.
0%DABstep benchmark accuracy
Visit Energent.ai
07 — Beyond work
Curious about everything.
- PlacesSan Francisco ⇄ Abu Dhabi, by way of Waterloo.
- ClimateVolunteer with The Climate Reality Project.
- EducationVolunteer with the Visionary Education Foundation.
- CredentialsAWS ML Specialty, AWS Solutions Architect, Google Cloud Data Engineering, SOA P & FM, FRM Level I.
