A disciplined and structured approach to learning and implementing the fundamentals of AIML.
through bleeding edge concepts, and an amazing peer group to learn with.
Through ERA, EVA, END and EPAi, TSAI has trained more than 7000 students!
Our ERA V2 course is a fusion of our most successful offerings: EVA (Extensive Vision AI), END (Extensive NLP via Deep Learning), key sessions from EMLO (Extensive Machine Learning Ops) and learnings from ERA V1. With an updated curriculum covering both vision and NLP topics as well as machine learning operations, you will not only learn the latest AI techniques but also gain practical experience in developing applications and deploying them on the cloud.
Immerse yourself in the world of advanced AI as you master and train state-of-the-art models such as ResNet, YOLO, BERT, CLIP, Stable Diffusion, RL & ChatGPT.
Registrations are opening on 15th December! Classes would start from 20th Jan 2024
STAGE #1
| Lecture Title |
|---|
| Fundamentals of Artificial Intelligence |
| Exploring Neural Network Architectures |
| Git and Python Essentials |
| Building the First Neural Networks |
| Introduction to PyTorch |
| Backpropagation and Advanced Architectures |
| In-Depth Coding Practice |
| Advanced Techniques and Optimizations |
| Data Augmentation and Visualization |
| PyTorch Lightning and AI Application Development |
| Residual Connections in CNNs and FC Layers |
| Building and Deploying AI Applications |
| YOLO and Object Detection Techniques |
| Multi-GPU Training and Scalable Model Serving |
| UNETs, Variational AutoEncoders, and Applications |
Stage #2
| Lecture Title |
|---|
| UNETs, Variational AutoEncoders, and Applications |
| Transformers and Advanced Embedding Techniques |
| Encoder Architectures and BERT |
| Masked AutoEncoders and Vision Transformers |
| Decoders and Generative Pre-trained Transformers |
| Training and Fine-tuning Large Language Models |
| CLIP Models and Training |
| Generative Art and Stable Diffusion |
| Automatic Speech Recognition Fundamentals |
| Reinforcement Learning Part I |
| Reinforcement Learning Part II |
| Reinforcement Learning from Human Feedback |
| Training ChatGPT from Scratch |
| Training Multimodal GPTs |
| Capstone Project |
Download the full course syllabus here.
An advanced Python course for fundamental understanding of Python Language and the PyTorch library. Designed for those who want to become application and ML Architects.
This course is not for the beginners. Take this sample test to see what would you be learning.
Registrations will start on 15th December! Stay Tuned
PHASE #1 - FUNCTIONAL PYTHON
| Lecture Title |
|---|
| Basics: Python Type Hierarchy, Multi-line statements and strings, Variable Names, Conditionals, Functions, The While Loop, Break Continue and the Try Statement, The For Loop and Classes |
| Object Mutability and Interning: Variables and Memory References, Garbage Collection, Dynamic vs static Typing, Variable Re-assignment, Object Mutability, Variable Equality, Everything is an Object and Python Interning |
| Numeric Types I: Integers, Constructors, Bases, Rational Numbers, Floats, rounding, Coercing to Integers and equality |
| Numeric Types II: Decimals, Decimal Operations, Decimal Performance, Complex Numbers, Booleans, Boolean Precedence and Comparison Operators |
| Functional Parameters: Argument vs Parameter, Positional and keyword Arguments, Unpacking Iterables, Extended Unpacking, __*args_, Keyword Arguments, __**kwags_, Args and Kwargs together, Parameter Defaults and Application |
| First Class Functions Part I: Lambda Expressions, Lambdas and Sorting, Functional Introspection, Callables, Map, Filter, Zip and List Comprehension |
| First Class Functions Part II: List Comprehension, Reducing functions, Partial Functions, Operator Module, Docstrings and Annotations. |
| Scopes and Closures: Global and Local Scopes, Nonlocal scopes, Closures, and Closure Applications |
| Decorators: Decorators and Decorator applications (timers, logger, stacked decorators, memoization, decorator class and dispatching) |
| Tuples and Named Tuples: Tuples, Tuples as data structures, named Tuples, DocStrings, and Application |
| Modules, Packages and Namespaces: Module, Python Imports, importlib, import variants, reloading modules, __main__, packages, structuring, and namespaces |
| fStrings, Timing Functions and Command Line Arguments: Dictionary Ordering, kwargs, tuples, fStrings, Timing Functions and Command Line Arguments |
| Sequence Types I: Sequence Types, Mutable Sequence Types, List vs Tuples, Index Base and Slice Bounds, Copying Sequence and Slicing |
| Sequence Types II and Advanced List Comprehension: Custom Sequences, In-place Concatenation and Repetition, Sorting Sequences, List Comprehensions + Small Project |
| Iterables and Iterators: Iterating Collections, Iterators, Iterables, Cyclic Iterators, in-built Iterators, iter() function and iterator applications |
| Generators and Iteration Tools: Yielding and Generator Functions, Generator Expressions, Yield From, Aggregators, Chaining and Teeing, Zipping and their applications |
| Context Managers: Context Managers, Lasy Iterators, Generators and Context Managers, Nested Context Managers and their application |
| Data Pipelines: Data Pipeline and application |
Phase #2 - OOPS & PYTORCH
| Lecture Title |
|---|
| Hash Maps and Dictionaries: Associative Arrays, Hash Maps, Hash Functions, Dictionary Views, Handling Dictionaries and Custom Classes |
| Sets and Serialized Dictionaries: Set Theory, Python Sets, Frozen Sets, and Set Applications, DefaultDict, OrderedDict, Counters and UserDict |
| Serialization and Deserialization: Picking, JSON Serialization, Encoding and Decoding JSON, and Applications |
| Classes Part I: Object and Classes, Attributes, Callables, Functional Attributes and Run-time attributes |
| Classes Part II + DataClasses: Properties, Decorators, Read-Only Properties, Class and Static Methods, Scopes, Dataclasses and Application |
| Polymorphism and Special Methods: Polymorhpism, __str__ and __repr__ methods, rich comparisons, hashing and equality, callables, and applications |
| Single Inheritance: Single Inheritance, Object Class, Overriding, Extending, Delegation, Slots, and applications |
| Descriptors: Descriptors, Getters and Setters, Instance Properties, Strong and Weak References, __set_name__ method, Proprty Lookup Resolution and application |
| Enumerations and Exceptions: Enumerations, Aliases, Custom Enums, Python Exceptions, Handling and Raising Exceptions and creating custom exceptions |
| Pytorch Basics I : Matrices, Tensors, Variables, Numpy and PyTorch inter-operability, Rank, Axes and Shapes |
| PyTorch Basics II: Data and Dataloader, Forward Method, Training Loop and Training Pipeline |
| PyTorch Intermediate I + Pytorch Internals:PyTorch Classes, Containers, Layers and Activations. PyTorch Internals or how Pytorch uses Advanced Python internally |
| PyTorch Intermediate II: Distance and Basic Loss Functions, Utilities, Profiling Layers, MACs/FLOPs calculations and Memory Usage |
| PyTorch Advanced I: Convolution Algorithm Implementation, Autograd Mechanics and Dynamic Computation Graph |
| PyTorch Advanced II: Optimizers, Custom Dataloaders, Tensorboard Integration, Memory Management and Half Precision Training |
| PyTorch Advanced III: Advanced Loss Functions for GAN, Kullback Lieber, Embeddings, Focal, IoU, Perceptual, CTC, Triplet and DICE |
Course Feedback
| Feedback from Phase 1 students moving to Phase 2 |
|---|
| Initially i thought it would be just like all other python course, i joined just thinking i might learn few things more.. but as session progressed i was like.. okay i dont know python.. amazing sessions and course content |
| A very in depth course and excellent concepts |
| I really liked the course content. I never learnt python in this much depth. Now I can say that I am a python developer 😊 |
| Slightly hectic, when it comes to assignment submission.More days(atleast week time should be given for assignment submission) |
| ONE OF THE BEST COURSE AT AFFORDABLE PRICE |
| Has helped me know about intricate things on python |
| Course content is good and in depth which makes it easy to understand for anyone |
| Seriously EPAi one of the best Intermediate Python Courses that I have taken up. In-depth Content, Fun-To-Do Assignments, what else do I need? |
| Honestly i learned a lot from this course.. Course contents are really good and covered in depth which i really liked. |
| It is really awesome course !!! Good in-depth sessions. |
| For me course really helped a lot. Got to learn many new things. |
| Course is well structure but would request for more time for the assignment submissions. Like other courses a week should be good in my opinion. |
| Excellent stuff |
| Very nice course I learned a lot |
| The course content is very well structured and assignments are also top notch |
| Excellent course |
| This course is extremely unique. Not surprised as it's from TSAI. My overall experience with TSAI has been amazing ever since. |
| Hits the very core foundation of important concepts |
| One of the best courses I have ever attended |
| Good course content, too fast pace, challenging assignments |
| Gives in depth knowledge about the design and working behaviour of python |
| Advanced course in python which teaches CI/CD as an extra practice. |
| It's very good |
| Its great course for Advance Pythons for AI |
| good platform to explore so many new concepts |
| In-depth understanding of python for writing optimized, error-free and modular codes. Creating our own packages and module. |
| Course content is good and touching the deep roots of python |
| The course is very useful for me, I have not seen a course with this good content online. |
| its beautiful |
| Awesome, worth it |
| I think it's a very exhaustive course and is really going to be useful for my career. |
Instructor Feedback
| Feedback from Phase 1 students moving to Phase 2 |
|---|
| ABOUT THE INSTRUCTOR |
| I joined because you were teaching |
| Take a bow |
| What can I say, he is the best person to teach this course |
| Too good |
| BEST I CAN SAY. ROHAN SHRAVAN IS MY NEW ROLE MODEL. I WISH HIM SUCCESS FOR HIS FUTURE PLANS. |
| PRO |
| Rohan is an amazing instructor and makes every attempt to clarify things |
| G.O.D |
| One of the best instructor and influencer i have seen in my life. |
| Fabulous. Please take some more courses like this - C++, Javascript. |
| When i think of Rohan, I see dedication, commitment and discipline. when i look back i see much change in me. Thank you very much Rohan. |
| Good as always |
| Instructor is awesome like always! |
| Class apart |
| good |
| Knowledgeable |
| Excellent |
| Knowledgeable |
| Awesome |
| Awesome |
| Good hardworking |
| Perfect |
| Good |
| Extremely knowledgeable and experienced. Honoured to have such a mentor |
| Knowledgeable, explains even the tiniest detail |
| Rohan is very very very knowledgeable. And more than that, he is an awesome and inspirational teacher. He knows the issues that the students might face and keep his content and hands on session based on these factors. |
| Man! No comments! |
| Perfect to describe in one word and easily one of the best teachers I have studied under in my life yet |
| expert in python |
| Good |
| One of the finest instructor from whom I have taken the course |
| He gives great insight into topic taken |
| Excellent & Has lot of passion & patience |
| He is well knowledge in the topics |
| Rohan has ability to deliver complex concepts in a nice and simple yet powerful way. |
| Rohan is awesome. |
| Rohan is exceptional. |
| cant get better |
| I don't think it can get better than Rohan. Its his humble nature and passion that he brings to each class that drives me. |
A cutting-edge course for mastering the art of managing and deploying machine learning models at scale.
This course is not for the beginners.
EMLO
| Lecture Title |
|---|
| Introduction to MLOps; Best practices and tools for managing, deploying, and maintaining ML models. |
| Docker - I; Creating Docker containers from scratch and understanding core concepts. |
| Docker - II; Introduction to Docker Compose for deploying ML applications. |
| PyTorch Lightning and Project Setup; High-performance training and deployment using PyTorch Lightning. |
| Data Version Control (DVC); Managing ML data and models using DVC. |
| Experiment Tracking; Using Tensorboard, MLFlow, and Hydra for experiment tracking and configuration. |
| HyperParameter Optimization; Techniques using Optuna and Bayesian Optimization. |
| Deployment for Demos - Gradio; Creating and sharing ML model demos using Gradio. |
| AWS CrashCourse; Covering EC2, S3, ECS, ECR, and Fargate for ML model deployment. |
| Model Deployment with FastAPI; Deploying ML models using FastAPI. |
| Deploying CLIP; CLIP-as-a-Service with caching on Redis and gRPC endpoints. |
| Model Deployment on Serverless; Introduction to AWS Lambda and Google Cloud Functions. |
| Model Deployment with TorchServe; Deploying ML models using TorchServe. |
| Model Runtime in Browser; Running ML models in the browser. |
| Model Explainability; XAI techniques such as SHAP and LIME |
| Kubernetes - I; Introduction to Kubernetes and its key concepts. |
| Kubernetes - II; Monitoring and configuring Kubernetes clusters for ML workloads. |
| Deploying Models on a Cluster; Using KServe for deploying ML models on Kubernetes. |
| CI/CD Pipeline - I; Preprocessing, training, and model packaging in CI/CD pipelines. |
| CI/CD Pipeline - II; Model registration and deployment for staging and production. |
| CI/CD Pipeline - III; Connecting the pipeline to a code repository for auto-training and deployment. |
| Auto Scaling Deployment and Stress Testing; Techniques for auto-scaling and stress testing. |
| Model Monitoring and Alerting; Using Grafana and Prometheus for monitoring and alerting. |
| Fine Tuning and Deploying Stable Diffusion; Using DreamBooth for fine-tuning and deployment. |
| Model Performance Optimization; Techniques such as FP16 CUDA_HALF, INT8 Quantization, CUDA AMP, Tritd>Server, and TensorRT. |
| Fine Tuning and Deploying LLMs; Using FSDP, PEFT, and LoRA for fine-tuning and deployment. |
| Deploying Chat with LLM; Deploying a chat instruction-tuned LLM for conversation. |
| Connecting LLMs with External Data; Augmenting LLMs with external data using in-context learning and data indexing. |
| Capstone - I; Develop and deploy an end-to-end MLOps pipeline as a final project. |
| Capstone - II; Conclusion of the final project and course. |
Download the full course syllabus here.
EVA was the most exhaustive and updated Deep Vision Program in the world! It was spread over three semester-style phases, each restricted by a qualifying exam. It has now been merged into ERA.
Phase #1
| Fundamentals of DNNs & Transformes |
|---|
| 1: Background & Basics: Machine Learning Intuition |
| 1.5: Python: Python 101 for Machine Learning (Handson [HO] 1) |
| 2: Neural Architecture: Convolutions, Pooling Operations & Channels |
| 2.5: PyTorch & Lightning: PyTorch 101 for Vision Machine Learning [HO2] |
| 3: First Neural Network: Kernels, Activations, and Layers |
| 4: Architectural Basics: Building blocks of DNNs |
| 5: Coding Drill Down: We go through 9 model iterations together, step-by-step to find the final architecture [HO3] |
| 6: Mathematical Foundation: For ML & Backpropagation [HO4] |
| 7: Advanced Convolutions & Augmentation: Introduction to Advanced Conv Concepts & Albumentations LIbrary |
| 8: RFs & Attention: RFs & Attention is all you need! |
| 9: Advanced Training & LRs: Class Activation Maps, Optimizers, LR Schedules, LR Finder & One Cycle Policy |
| 10: Super Convergence: Training Models at lightning speed [HO5] |
| 11: Transformers : Deepdive Coding into Transformers [HO6] |
| 12: ViT: Vision Transformers |
| 13: YOLO Part 1: Object Detection (Data Collection and Processing) |
| 14: YOLO Part 2: Training an Object Detector [HO7] |
| 15: Capstone: Your turn now! Qualifying Exam for Phase 2 |
Phase #2
| Phase #2 - Transformers & Stable Diffusion |
|---|
| 16: Transformers and Attention Refresher (HandsOn 8) |
| 17: Attention and its Types |
| 18: Vision Transformers Part 1 |
| 19: Vision Transformers - ViT Part 2 (HandsOn 9) |
| 20: Self-Distillation and Self-Supervised Vision Transformers |
| 21: Introduction to Embedding and Language Models |
| 22: Language Model via Transformers (HandsOn 10) |
| 23: Advanced Concepts in Training Transformers (HandsOn 11) |
| 24: Semantic Segmentation |
| 25: Generative Adversarial Networks (HandsOn 12) |
| 26: Variational AutoEncoders & Mathematics |
| 27: VAE, its types, and Training (HandsOn 13) |
| 28: CLIP & Other Advanced Training Concepts (HandsOn 14) |
| 29: Generative AI and Stable Diffusion |
| 30: Capstone Project |
Phase #3
| Phase #3 - Applications |
|---|
| 31: Model Quantization & Training Part 1 |
| 32: Model Quantization & Training Part 2 |
| 33: Optical Character Recognition |
| 34: Depth Estimation |
| 35: Image Deblurring, Denoising & Enhancement |
| 36: Image Super Resolution |
| 37: Video Super Resolution |
| 38: Pose Estimation |
| 39: Face Recognition |
| 40: Understanding Audio for DNNs |
| 41: Audio Classification & Noise Removal |
| 42: Speech Enhancement & Separation |
| 43: Automatic Speech Recognition |
| 44: Beam Search for STT & NLP |
| 45: Capstone Project Speech Stable Diffusion! |
| Bottom Topics will soon be moved to another course on Reinforcement learning |
| Reinforcement Learning Basics: Markov Decision Processes, Deterministic, and Stochastic Environments & Bellman Equation |
| Q-Learning: Q-Learning, Plan vs Policy Networks, and Environment Models |
| Deep Q-Learning & DeepTraffic: Custom Environments, OpenGym, Exploration vs Exploitation, and improvements to DQN |
| Deep Reinforcement Learning: Policy Gradients, Dynamic Programming, Policy Evaluations, and Temporal Difference Learning |
| Actor-Critic Models: Memory Structures, Gibbs Softmax, Eligibility Traces, and Polyak Averaging |
| A3C Models: A3C, A3C optimizations, and implementation logic |
| Deep Deterministic Policy Gradients: DDPG Background, Off-Policy Networks, Continuous Action Spaces, and Replay Buffers |
| Twin Delayed DDPG Part 1: Clipped Double-Q Learning, Delayed Policy Updates, and Target Policy Smoothing |
| Twin Delayed DDPG Part 2: Full TD3 implementation to make a robot walk, and solve a custom environment |
| Autonomous Robotics Introduction: Introduction to ARI platform, and its control systems. Real Robot we mean! |
| Sensor Fusion for Localization: Sensor fusion, depth estimation, and stereo imaging for robotic localization |
| 3D Environmental Reconstruction Part 1: Solving 3D mapping for static environment |
| 3D Environmental Reconstruction Part 2: Solving 3D mapping for dynamic (moving) objects in the static environment |
| 3D Environmental Reconstruction Part 3: Solving 3D mapping for dynamic objects in a dynamic environment |
| Advanced Path planning, and Navigation: A*, and other Path planning, and algorithms |
| EndGame: CapStone project to implement everything we learned |
| The later part of this course's topics are inspired from Udacity Nanodegree but only just the topics, not its contents. We would be implementing these on a real robot, without ROS, and using TD3, not DQN as in Udacity |
In our flagship NLP program, we had ditch RNN/LSTMS and other recurrent networks completely, and focused fully on Transformers. After covering the basics of Neural Networks, we focused on Attention is All you need then covering advanced transformers like BERT, BART, ending with Retrieval Augmented Generation.
The world since last year moved on, and so has END. END is now merged with ERA.
PHASE #1 - Transformers
| Lecture Title |
|---|
| Background and Basics of Modern NLP |
| From Embeddings to Language Models |
| Advanced Python for General Computing & NLP |
| PyTorch for NLP |
| RNNs are dead and their Renewed Relevance |
| GRUs, Seq2Seq and Attention Mechanism |
| HandsOn Training 1 |
| Deep NLP using Convolutions |
| HandsOn Training 2 |
| Attention and Memory in Deep NLP |
| HandsOn Training 3 |
| Transformers with Linear Attention |
| HandsOn Training 4 |
| GloVe, Memory Networks and Recap |
| Infinity Capstone Project |
Phase #2 - GPT
| Lecture Title |
|---|
| Transformers and Attention Mechanism - Overview |
| Reformer: the efficient transformer |
| Bi-Directional Transformers |
| Document Level Models & Contextual Representations |
| GPT1 & Models of Dialog |
| GPT1 Coding & Practice |
| Building and training GPT2 and BERT |
| GPT2 and BERT Coding & Practice |
| GPT3 Deep Dive: Part 1 Architecture & Preprocessing |
| GPT3 Deep Dive: Part 2 Training |
| GPT3 Coding & Practice 1 |
| GPT3 Coding & Practice 2 |
| GPT3 Coding & Practice 3 |
| Advanced NLP Over the Edge |
| Endgame Capstone Project |