Predictive model on a real tabular dataset
Deliverable: A trained and tuned model with a written evaluation covering metric choice, baseline comparison, cross-validation results and a documented analysis of where the model fails.
Course
Artificial Intelligence
Python, machine learning foundations and applied generative AI in one track — ending with a retrieval-augmented application you have built, evaluated and can explain end to end.
- Duration: 6 months
- Classroom · Online live
- Level: intermediate
What you will be able to do
- Train, tune and evaluate supervised machine learning models on real tabular datasets
- Explain why a model made a given prediction and where it is likely to fail
- Build production-shaped prompts with evaluation criteria rather than trial and error
- Design and build a retrieval-augmented generation system over a private document set
- Measure an AI system's output quality with a defined test set instead of impressions
- Recognise the failure modes of generative models and design guardrails around them
Who this course is for
- Graduates in engineering, science, mathematics, statistics or computer applications
- Working developers, analysts or testers moving toward AI roles
- Comfort with basic programming logic; we cover Python from fundamentals but move quickly
- Class 12 mathematics is sufficient — we teach the linear algebra and statistics you need
Syllabus
8 modules · 6 months
Module 1. Python and the mathematics you actually need
- Python for data work — types, comprehensions, functions and modules
- NumPy arrays and vectorised thinking
- Vectors, matrices and dot products, taught geometrically
- Derivatives and gradients as "which way is downhill"
- Probability, distributions and Bayes in practical terms
Module 2. Machine learning foundations
- Framing a problem as supervised, unsupervised or neither
- Linear and logistic regression from first principles
- Decision trees, random forests and gradient boosting
- Train, validation and test splits, and how leakage ruins results
- Cross-validation, hyperparameter search and overfitting
- Choosing metrics — accuracy, precision, recall, F1, ROC-AUC and when each misleads
Module 3. Deep learning essentials
- Neural networks as stacked transformations
- Backpropagation and gradient descent in practice
- Training loops, batching and learning rate schedules
- Convolutional networks for images
- Transfer learning and why it beats training from scratch
Module 4. Language models and transformers
- Tokenisation, embeddings and what a vector actually represents
- Attention and the transformer architecture
- Pretraining, fine-tuning and instruction tuning
- Running open models locally with Hugging Face
- Parameter-efficient fine-tuning with LoRA
Module 5. Applied generative AI and prompt engineering
- Prompt structure — instruction, context, examples and output contract
- Few-shot prompting and chain-of-thought patterns
- Building an evaluation set before you start iterating
- Structured output and function calling
- Cost, latency and model selection trade-offs
Module 6. Retrieval-augmented generation
- Why RAG exists and what it does not solve
- Chunking strategies and their effect on retrieval quality
- Embeddings, vector stores and similarity search
- Reranking and hybrid retrieval
- Measuring groundedness and detecting hallucination
Module 7. Responsible and production AI
- Bias in training data and in evaluation sets
- Privacy, data residency and what not to send to a hosted model
- Guardrails, refusals and prompt injection
- Monitoring an AI feature after release
Module 8. Capstone and interview preparation
- Scoping a capstone that can be finished and defended
- Code review and refactoring for a portfolio repository
- AI and ML interview question practice
- Explaining trade-offs under questioning
Tools and technologies you will use
- Python
- NumPy
- pandas
- scikit-learn
- PyTorch
- Hugging Face Transformers
- LangChain
- Vector databases (FAISS and Chroma)
- Git
- Streamlit
Projects you will build
Image classifier using transfer learning
Deliverable: A fine-tuned vision model with a confusion matrix, per-class error analysis and a short note on which misclassifications would matter in deployment.
Prompt evaluation harness
Deliverable: A test set of at least fifty cases with scoring criteria, plus a comparison of three prompt strategies showing measured quality differences rather than anecdotes.
Retrieval-augmented assistant over a private corpus
Deliverable: A working RAG application with a documented chunking and retrieval strategy, a groundedness evaluation, and an honest write-up of the queries it still answers badly.
Where this course can take you
- AI Engineer
- Machine Learning Engineer
- Applied AI Developer
- Data Scientist
- AI Solutions Analyst
Duration, modes and fees
- Duration
- 6 months
- Delivery modes
- Classroom · Online live
- Fees
- Share your details for the current fee
- Fees vary by batch and delivery mode. Share your details and an advisor will confirm the current fee.
Placement assistance
Every student gets placement assistance — that is what 100% placement assistance means. It is support for all, not a job for all. We do not promise a specific salary, a specific number of interviews, or placement at any named company, and you should be wary of anyone who does.
What is included
- A place in the monthly placement drive, held every third Saturday
- The readiness programme every second Saturday — mock interviews and preparation
- CV review against the specific roles you are targeting
- Portfolio review, so your project work is presented the way a reviewer will read it
- Access to the vacancy pool employers send directly to the Samyak network
- Guidance on which roles realistically fit your background and which do not
- A place in the next drive, with coaching, if you are not selected in this one
What is not included
- Any guarantee of a job, an interview, or a particular salary
- Placement at a named or partner company
- Applying to jobs on your behalf
- Support before you have completed the course and its project work
- Visa, relocation or overseas placement assistance
What this course refuses to do
It does not open with a chatbot demo. Building something impressive-looking with an API key takes an afternoon and teaches almost nothing transferable, and the market is now full of people who can do exactly that and no more.
The differentiator in an AI role is judgement: knowing why a model is wrong, whether a benchmark number means anything, when retrieval will help and when it will not, and how to prove that a change improved things. Judgement is built on foundations, so the first half of this course is foundations.
The evaluation habit
Every applied module in the second half requires you to build a test set before you build the feature. This is deliberate and it is the single most employable habit in the course. Most people iterate on prompts by feel and declare victory when an output looks good. You will be able to say a change improved groundedness on a fifty-case set — which is the difference between an opinion and a result.
Honest limits
Six months gets you to the point where you can build, evaluate and reason about applied AI systems. It does not make you a research engineer, and it does not replace a strong software engineering foundation if you want to work on production AI infrastructure. We say so on the first day.
Questions
Artificial Intelligence — frequently asked questions
Do I need a mathematics degree to take an AI course?
No. You need Class 12 mathematics and a willingness to work through the first module. We teach linear algebra, gradients and probability specifically as they are used in machine learning, geometrically rather than as proofs, which is enough for everything that follows.
Does this course cover generative AI or only traditional machine learning?
Both, in that order. Modules two and three build machine learning and deep learning foundations, and modules four to six cover transformers, prompt engineering and retrieval-augmented generation. The foundations come first because debugging a generative system without them is guesswork.
Will I learn to build my own large language model?
No, and no six-month course honestly can — pretraining a competitive model costs millions of dollars. You will learn to fine-tune open models, build retrieval systems around them, and evaluate their output rigorously, which is what applied AI roles actually involve.
How is this different from your data science course?
Data science centres on statistical modelling and drawing conclusions from data. This course goes further into deep learning and generative systems, and spends significant time on building and evaluating AI applications rather than only analysing data.
What kind of laptop do I need for this course?
Any machine with 8GB RAM is enough for most of the course, because we use hosted notebooks with free GPU access for the heavier deep learning work. For the local model modules, 16GB RAM makes the experience considerably smoother but is not mandatory.
Enquire about Artificial Intelligence
Three details is all we need. A course advisor will call you back.
Related courses
- Data Analytics
A structured path from spreadsheets to SQL, Power BI and Python, built around the four things a data analyst is actually paid to do — pull the data, clean it, analyse it and explain it.
- Full Stack Development
Front end, back end, databases and deployment taught as one connected system, ending with applications you have shipped to a public URL and can walk an interviewer through.
Next step
Talk to a course advisor
Tell us what you want to learn and we will help you pick the right course, batch and mode.