Lab 01
Neural Network from Scratch (NumPy only)
Write a 2-layer network using only NumPy. Manually implement the forward loop, analytical backprop and gradient updates — no autograd.
Become a Deep Learning engineer in 14 weeks.
Neural networks written from scratch in NumPy. Modern CNNs, LSTMs, Transformers, VAEs, GANs and Diffusion — all in PyTorch. Ship 6+ portfolio models, mentored by working ML engineers.
0% EMI availablePlacement guaranteeHands-on labs included
Deep Learning Engineer Track
30-min call · no pressure · get a personal roadmap
14 wks
Cohort length
60+
PyTorch labs
9.5/10
Alumni rating
90%
Placement rate
Alumni working at
FlipkartFreshworksRazorpayMeeshoCredNvidiaPostmanZomatoGrowwOla
Outcomes
What you'll be able to do on day one at your new job
Every module ends with a graded lab that mirrors what teams actually ship. No slides-only theory.
Derive forward + backprop by hand and implement a network in pure NumPy.
Debug vanishing/exploding gradients with Xavier/He init and BatchNorm.
Train a ResNet image classifier with transfer learning and augmentation.
Detect objects using bounding boxes, IoU and the YOLO family.
Build LSTM / GRU sequence models for text and time-series data.
Explain Multi-Head Self-Attention and Transformer positional encoding.
Compare GPT (decoder-only) vs BERT (encoder-only) LLM paradigms.
Train VAEs, GANs and reason about Diffusion model intuition.
14-week curriculum
Get the detailed syllabus PDFSix modules. Foundations to modern architectures.
- Module 01Weeks 1–3
Foundations of Neural Networks
Build every block from single neurons to full MLPs.
- Linear algebra & calculus review
- Perceptron, bias & activation functions
- Sigmoid, Tanh, ReLU, Leaky ReLU
- Multi-Layer Perceptrons (MLPs)
- MSE, BCE, Categorical Cross-Entropy
- Backpropagation vectorization with NumPy
- Module 02Weeks 4–5
Optimization, Regularization & Training Mechanics
Stop deep networks from silently failing.
- SGD, Batch GD, Mini-batch GD
- Momentum, RMSprop, Adam
- Vanishing / exploding gradients
- Xavier & He weight initialization
- L1/L2 regularization, Dropout, early stopping
- Batch Normalization & Layer Normalization
- Module 03Weeks 6–8
Computer Vision (Convolutional Neural Networks)
Grid-structured data without exploding parameter counts.
- Convolutions, filters, padding, strides
- Max & Average Pooling
- LeNet, AlexNet, VGG, ResNet
- Transfer learning & fine-tuning
- Data augmentation strategies
- Object detection: bounding boxes, IoU, YOLO
- Module 04Weeks 9–10
Sequence Modeling & NLP
Order-dependent, temporal data.
- RNNs, hidden states, unrolling
- Vanishing memory problems
- LSTM: forget/input/output gates
- GRU architectures
- Seq2Seq Encoder-Decoder pipelines
- Module 05Weeks 11–13
Modern Transformer Architectures & LLMs
The backbone of modern Generative AI.
- Scaled Dot-Product Attention
- QKV matrices & Multi-Head Attention
- Positional encoding & masked attention
- Encoder-Decoder Transformer stack
- GPT (decoder-only) vs BERT (encoder-only)
- Tokenization pipelines & scaling laws
- Module 06Week 14
Unsupervised & Deep Generative Models
Compress, represent and synthesize new data.
- Autoencoders & denoising
- Variational Autoencoders (VAEs) & KL divergence
- Generative Adversarial Networks (GANs)
- Diffusion model intuition
Mandatory capstone labs
Projects you'll actually ship
Lab 02
ResNet Image Classifier with Transfer Learning
Load a pre-trained ResNet in PyTorch, swap the classification head, apply intense augmentation, and fine-tune on a custom medical or environmental dataset.
Lab 03
Character-Level Text Generator
LSTM sequence model that ingests text files, predicts the next character, and generates new passages tuned via temperature sampling.
Experienced faculty
Learn from engineers who've operated this at scale
Every instructor has 8+ years in production roles at companies whose logos you already know.
D
Dr. Rajesh Verma
Lead Instructor · ex-Nvidia
PhD in ML, 12 yrs building vision models. Contributed to internal PyTorch training kernels.
S
Shreya Kapoor
NLP Faculty · ex-Freshworks
Built LLM-powered support automation for 5M+ tickets. Speaker at PyData Bangalore.
K
Kunal Bhatia
Vision Mentor · ex-Ola
Trained the object detection stack behind Ola's driver safety product. 8 yrs in production CV.
800+
ML alumni placed
150+
Hiring partners
8 yrs
Training experience
12
Max cohort size
Student stories
Real placements. Real hikes. Real portfolios.
"Writing backprop by hand once made every framework feel obvious. Best decision of my ML journey."
A
Aditi Rao
ML Engineer · SaaS unicorn
"The Transformer module was the sharpest I've seen — I could actually explain attention in interviews, not just recite it."
K
Karan Malhotra
NLP Engineer · fintech
"I built a working ResNet classifier for medical scans as my capstone. It became the first thing recruiters clicked on."
F
Fatima Sheikh
Computer Vision Engineer
"Mentors pushed me on math depth — that's what separated my interviews from the 100 other candidates."
R
Rohan Menon
Applied Scientist
FAQ
Everything parents & partners ask
Do I need a math background?
Comfort with matrix operations, derivatives and basic probability is enough. We do a 1-week review at the start. Half our alumni come from engineering, not pure math.
Do I need prior ML exposure?
Basic supervised ML (train/test split, overfitting, one classifier) helps. If you're new to ML, do a short intro course first — or ask us to include a 2-week bridge module.
Do I get GPU access for training?
Yes. Sandbox credits on Colab Pro + a Databricks / Vertex AI trial for larger runs. Every lab is designed to fit within provided credits.
Why PyTorch and not TensorFlow?
PyTorch dominates research and modern production ML teams in India. Skills transfer to TF/Keras in a week if a specific role needs it.
What roles do alumni get placed into?
ML Engineer, Applied Scientist, NLP Engineer, Computer Vision Engineer and Deep Learning Researcher — with typical packages of ₹14-32 LPA.
What's the fee and are EMIs available?
Fee is ₹89,000 all-inclusive. 0% interest EMIs across 6, 9 and 12 months. Scholarships available for women returnees and tier-3 students.
Next cohort · Aug intake
Your Data Engineering career starts with a 30-min call.
Talk to a mentor. Get an honest read on your background, a personalized roadmap, and full fee + EMI details. No pressure.