Théodore Wajnberg
Machine Learning Research
Focuses on model interpretability. Teaches modules 1 and 2. Has published 9 papers on neural network behavior in low-data regimes.
This program covers how AI systems are actually built - the data, the models, the tradeoffs. It is designed for people with no machine learning background who want a technically honest starting point.
The sequence matters: each module introduces concepts used in the next. By session 8, participants are reading and discussing real research papers because they have the vocabulary to do so.
The program runs as live webinars twice a week, with a dedicated Q&A block in each session. Experts answer questions directly rather than through a forum thread answered three days later.
Participants who complete all 6 modules and submit a final project receive a digital certificate. Roughly 71% of people who start the program finish it, higher than typical online course completion rates, partly because the live format creates accountability.
Training, loss functions, overfitting. What the model is doing when it "learns" - explained without skipping the math.
Layers, activations, backpropagation. Why depth matters and where it stops helping.
Tokenization, embeddings, transformers. How a model reads a sentence - and why that framing is already a simplification.
Convolutional networks, image classification, object detection. What the model sees versus what you think it sees.
Bias, fairness metrics, and the gap between benchmark performance and real-world deployment. Specific cases, not abstractions.
Data prep, model selection, evaluation, iteration. A guided project using a real dataset - participants choose the problem.
Core theory explained at a level where you can read documentation without a translator.
Every module includes a hands-on exercise using Python notebooks - no setup required on your end.
Live Q&A in each session, with questions answered by the instructor rather than a teaching assistant reading from notes.
2 sessions per week keeps momentum without requiring more than 4–5 hours of total time weekly.
The 4 instructors work full-time in AI research or applied engineering. Teaching is not their primary job.
Machine Learning Research
Focuses on model interpretability. Teaches modules 1 and 2. Has published 9 papers on neural network behavior in low-data regimes.
NLP and Language Models
Works on multilingual transformer fine-tuning. Leads module 3. Particularly good at explaining why tokenization decisions matter in practice.
Computer Vision Engineering
Builds vision systems for industrial inspection. Teaches module 4. Brings real deployment constraints into every session, not just benchmark results.
AI Ethics and Applied AI
Advises organizations on responsible AI deployment. Teaches modules 5 and 6. Known for making the ethics module the one participants cite most often afterward.
Do I need a programming background to join?
No prior coding experience is required. Python basics are introduced in the first 2 sessions, and all notebooks are pre-configured so there's no environment setup on your end.
How many live sessions per week?
Two sessions per week, each running approximately 75 minutes. All sessions are recorded and available for replay within 24 hours of broadcast.
What if I miss a live broadcast?
Every session is recorded. You can submit questions in advance and receive written answers within 48 hours through the program's messaging system.
Is there a certificate at the end?
Participants who complete all 6 modules and submit the final project receive a digital certificate from istoclaide confirming completion of the AI Fundamentals program.
16 live sessions, all recordings, instructor Q&A, hands-on notebooks, and final project review included.