The era of data, taught as an engineering discipline. A three-month, 120-hour programme covering Python, statistics, supervised machine learning, model failure modes and LLM applications, taught at BeeTees Academy in Edappally, Kochi.
DSML assumes Python basics, SQL and a working data model already exist. That is why entry is gated: either you have completed DABI, or you clear an equivalent placement test.
Most 120-hour data science courses spend half their time re-teaching fundamentals and stop at “we covered machine learning.” Because DSML starts from a known floor, all 120 hours go into work that reaches a shippable outcome — a model you built, broke, fixed and defended, and a working retrieval-augmented application over a real document set.
Together with DABI this reads as a clean 3 + 3 ladder: six months from spreadsheet to deployed model.
Module 4 is the one that separates this from a YouTube playlist. Train/test discipline, data leakage, class imbalance, drift, and knowing when a model should not be built at all. It is taught by people who have watched models fail in production, because nobody who has not can teach why they did.
Twelve weeks, 120 hours, one capstone defended to a panel.
Functions, environments, version control, and the difference between a notebook and a script that someone else can run.
Distributions, sampling, correlation versus causation, confidence, and the tests people routinely misuse — taught as why this result does not mean what the chart implies.
Regression, classification, trees and boosting, with feature engineering treated as the part that actually decides the outcome.
Find the leakage, the imbalance and the drift in someone else's model and document exactly how it fails.
Embeddings, RAG, an agent that calls tools, evaluating non-deterministic output, and cost and latency as design constraints — a working app over a supplied document set.
Package, serve and monitor. A deployed endpoint or Streamlit app, not a notebook on your laptop.
Twelve weeks, 15 to 20 seats. Entry gate: DABI completion or an equivalent placement test.
| Module | Hours | What is covered |
|---|---|---|
| 1 · Python, properly | 15 | Functions, environments, version control, notebooks versus scripts. Picks up where DABI's light Python module stopped. |
| 2 · Statistics that decisions rest on | 15 | Distributions, sampling, correlation versus causation, confidence, and the tests people misuse. |
| 3 · Supervised machine learning | 25 | Regression, classification, trees and boosting. Feature engineering as the part that actually matters. |
| 4 · Evaluation & failure modes | 20 | Train/test discipline, leakage, imbalance, drift, and when a model should not be built at all. |
| 5 · LLM applications | 25 | Embeddings, RAG, an agent that calls tools, evaluating non-deterministic output, cost and latency as design constraints. |
| 6 · Ship it | 10 | Package, serve, monitor. Awareness rather than full MLOps. |
| 7 · Capstone + defence | 10 | One project of your choosing, defended to a panel. |
Entry gate: complete DABI, or clear a placement test covering Python basics, SQL and data modelling. The gate is not gatekeeping — it is what lets 120 hours reach a shippable outcome instead of stopping at an overview.
Who it is not for: Complete beginners, and commerce graduates with no coding background. Start with DABI — it is the intended route in, and the stack is cheaper and works better than a standalone start.
DSML is a three-month, 120-hour Data Science and Machine Learning programme at BeeTees Academy of Commerce, Edappally, Kochi. It covers Python, applied statistics, supervised machine learning, model evaluation and failure modes, LLM and RAG applications, deployment, and a defended capstone.
Either completion of the DABI programme at BeeTees, or an equivalent placement test covering Python basics, SQL and data modelling. The gate exists so that all 120 hours go into machine learning and LLM work rather than re-teaching fundamentals.
Yes, if you clear the placement test. Most commerce graduates will not, and are better served starting with DABI — the two stack as a 3 + 3 ladder and the combined route costs less and finishes stronger than a standalone start.
Yes — a full 25-hour module on LLM applications covering embeddings, retrieval-augmented generation, agents that call tools, evaluating non-deterministic output, and cost and latency as design constraints. You finish with a working RAG application over a supplied document set.
Fee and batch dates are being finalised and are shared with our enquiry list first. Call +91 80757 34949 or use the form on this page and a counsellor will contact you within 24 hours.
Fees and batch dates go to our enquiry list first. Leave your details and an academic counsellor will call you within 24 hours.
Four AI-era programmes launched together. They are designed to stack.