💼 Job Opportunities
Data Scientist, ML Engineer, AI Engineer, Applied Scientist, MLOps Engineer, LLM Engineer, Data Analyst at Google, Microsoft, Amazon, Meta, Flipkart, Swiggy, Zomato, PhonePe, Razorpay, Cred, Zerodha, Fractal, Mu Sigma, LatentView, Tiger Analytics, TCS, Infosys, Wipro, Deloitte, Accenture, EY, KPMG, McKinsey QuantumBlack, BCG Gamma, Bain Vector. Expected CTC: 6-18 LPA for freshers; 20-60 LPA for 2-4 years experience.
📚 Curriculum (10 Units)
Unit 1: Python for Data Science + NumPy/Pandas Foundations 7 topics
Python Essentials for Data Science
Lists, dicts, comprehensions, generators, lambdas, decorators, file I/O. Setting up Anaconda + Jupyter + VS Code.
NumPy: Arrays, Broadcasting & Vectorization
ndarray internals, broadcasting rules, slicing, fancy indexing, performance vs Python lists (100× speedups).
Pandas Series & DataFrame Basics
Creating from dicts/lists/CSVs, indexing with loc/iloc, filtering, sorting, dropna, fillna, astype.
🔒 Enroll to unlock
Pandas GroupBy, Merge & Pivot
Split-apply-combine pattern, joins (inner/outer/left/right), pivot_table, melt, time-series resampling.
🔒 Enroll to unlock
Reading Real Data: CSV, Excel, JSON, SQL, Parquet
pd.read_*, SQLAlchemy connections, handling Indian datasets (GSTN, MCA, RBI CSVs).
🔒 Enroll to unlock
Project: UPI Transaction Analysis
Clean 10M-row NPCI-format UPI dataset, compute monthly trends, detect anomalies. Mini-project delivered in Jupyter.
🔒 Enroll to unlock
Performance: Vectorization vs Apply vs Loops
Why apply() is 10× slower than vectorization; when to use numba/Cython; measuring with %%timeit.
🔒 Enroll to unlock
Unit 2: Data Visualization, EDA & Feature Engineering 7 topics
Matplotlib + Seaborn Fundamentals
Figure/Axes model, subplots, saving high-DPI for reports. Histograms, box, violin, heatmap, pairplot.
🔒 Enroll to unlock
Plotly & Interactive Dashboards
Plotly Express for rapid interactive charts, Dash for internal dashboards, Streamlit for data apps.
🔒 Enroll to unlock
Exploratory Data Analysis (EDA) Framework
Profile → check missingness → univariate → bivariate → multivariate → target correlation → leakage check.
🔒 Enroll to unlock
Handling Missing Data & Outliers
MCAR/MAR/MNAR, imputation (mean/median/KNN/iterative), outlier detection with IQR and Isolation Forest.
🔒 Enroll to unlock
Feature Engineering for Tabular Data
One-hot, target encoding, frequency encoding, binning, log/box-cox transforms, date-time decomposition.
🔒 Enroll to unlock
Feature Scaling, Selection & Dimensionality Reduction
StandardScaler vs MinMaxScaler vs RobustScaler, chi², mutual info, PCA, t-SNE, UMAP.
🔒 Enroll to unlock
Project: EDA on Indian Stock Market (NSE)
Pull 5-year Nifty50 data via yfinance; volatility analysis, sector heatmap, rolling correlation.
🔒 Enroll to unlock
Unit 3: Statistics, Probability & Hypothesis Testing 7 topics
Descriptive Statistics & Distributions
Mean/median/mode, variance, skewness, kurtosis. Normal, binomial, Poisson, exponential with scipy.stats.
🔒 Enroll to unlock
Central Limit Theorem & Sampling
Why sampling distributions matter, bootstrap, standard error, confidence intervals.
🔒 Enroll to unlock
Hypothesis Testing (Z, t, Chi², ANOVA)
Null/alternative, p-value interpretation (NOT "probability null is true"), type I/II errors, power.
🔒 Enroll to unlock
A/B Testing for Data Science Interviews
Sample size calculation, MDE, stratification, Simpson's paradox — real product-data-science questions.
🔒 Enroll to unlock
Bayesian Thinking Primer
Bayes' rule, priors/posteriors, MAP vs MLE, why Bayesian matters for small-data medical / fraud problems.
🔒 Enroll to unlock
Correlation vs Causation & Confounders
Pearson vs Spearman, confounding, do-calculus intro, instrumental variables in a nutshell.
🔒 Enroll to unlock
Project: Cricket IPL Match Analysis
Test whether toss win is statistically correlated with match win over 2008-2025 IPL data.
🔒 Enroll to unlock
Unit 4: Classical Machine Learning (Supervised & Unsupervised) 7 topics
Linear & Logistic Regression from Scratch
Closed-form + gradient descent. Assumptions, multicollinearity, regularization (L1/L2/ElasticNet).
🔒 Enroll to unlock
Decision Trees, Random Forests, Gradient Boosting
Gini/entropy, tree depth, feature importance. XGBoost, LightGBM, CatBoost practical comparison.
🔒 Enroll to unlock
SVM, KNN, Naive Bayes
Kernel trick (RBF, polynomial), curse of dimensionality in KNN, Laplace smoothing in NB.
🔒 Enroll to unlock
Clustering: KMeans, DBSCAN, Hierarchical
Elbow method, silhouette, when DBSCAN beats KMeans (non-convex clusters).
🔒 Enroll to unlock
Cross-Validation, Hyperparameter Tuning
K-fold, stratified K-fold, time-series CV. GridSearchCV vs RandomizedSearchCV vs Optuna.
🔒 Enroll to unlock
Evaluation Metrics Deep Dive
Accuracy vs precision/recall/F1, ROC-AUC vs PR-AUC, log-loss, MAE vs MSE vs MAPE, quantile loss.
🔒 Enroll to unlock
Project: Credit Risk Scoring for Indian Fintech
Build XGBoost model on synthetic CIBIL-style features; discuss fairness and interpretability (SHAP).
🔒 Enroll to unlock
Unit 5: Deep Learning (CNN, RNN, Transformers) 7 topics
Neural Networks Fundamentals + PyTorch Setup
Perceptron, backprop intuition, activation functions, initialization. PyTorch tensors and autograd.
🔒 Enroll to unlock
CNNs for Computer Vision
Convolution, pooling, ResNet/EfficientNet/ConvNeXt. Transfer learning with torchvision.
🔒 Enroll to unlock
RNNs, LSTMs & GRUs
Sequence modeling, vanishing gradients, when NOT to use RNNs anymore (hint: usually).
🔒 Enroll to unlock
Transformer Architecture Explained
Self-attention, multi-head, positional encoding, encoder-decoder. The "Attention Is All You Need" paper walkthrough.
🔒 Enroll to unlock
Fine-tuning Pretrained Models
HuggingFace Transformers library, LoRA, QLoRA, PEFT for cheap fine-tuning on consumer GPUs.
🔒 Enroll to unlock
Regularization & Optimizer Deep Dive
Dropout, BatchNorm vs LayerNorm, Adam/AdamW vs SGD+momentum, learning-rate schedules.
🔒 Enroll to unlock
Project: Fine-Tune a Vision Model on Indian Cuisine Images
Collect dataset from web, fine-tune EfficientNet to classify 50 Indian dishes. Deploy to Streamlit.
🔒 Enroll to unlock
Unit 6: Natural Language Processing & Embeddings 7 topics
Text Preprocessing Pipeline
Tokenization (word/BPE), stemming vs lemmatization, stopwords (multilingual), normalization for Hindi/Telugu.
🔒 Enroll to unlock
Classical NLP: TF-IDF + Logistic Regression Baseline
Still beats BERT on some short-text classification! Build sentiment classifier on Twitter India data.
🔒 Enroll to unlock
Word Embeddings: Word2Vec, GloVe, FastText
CBOW vs skip-gram, subword embeddings for Indian languages.
🔒 Enroll to unlock
Contextual Embeddings: BERT, RoBERTa, IndicBERT
AI4Bharat IndicBERT for Indian languages, mBERT for multilingual, when contextual beats static.
🔒 Enroll to unlock
Sentence Embeddings for Semantic Search
Sentence-BERT, all-MiniLM-L6-v2, cosine similarity, building a "find similar questions" service.
🔒 Enroll to unlock
Named Entity Recognition + Relation Extraction
spaCy + custom NER for invoices, resumes. Distant supervision for relation extraction.
🔒 Enroll to unlock
Project: Multilingual Sentiment on Indian E-commerce Reviews
Amazon/Flipkart reviews in English+Hindi+Tamil; IndicBERT fine-tune with HuggingFace Trainer.
🔒 Enroll to unlock
Unit 7: LLM Integration (Claude, GPT-4o, Gemini APIs) 7 topics
LLM Landscape 2026: Claude, GPT, Gemini, Open-Source
Model comparison, context windows, pricing, latency, when to use which. Open-source: Llama 3.3, Mistral, Qwen.
🔒 Enroll to unlock
Claude API Deep Dive (Opus, Sonnet, Haiku)
Messages API, streaming, tool use, structured outputs, vision. Rate limits and cost optimization.
🔒 Enroll to unlock
OpenAI GPT-4o + o-series API
Completions vs Chat, function calling, reasoning models (o1/o3), streaming, JSON mode.
🔒 Enroll to unlock
Google Gemini 2.x + Vertex AI
Multimodal input (text, image, video, audio), Gemini Live, grounding with Google Search, Vertex deployment.
🔒 Enroll to unlock
OpenRouter + LiteLLM for Model Switching
One API, many providers. Fallbacks, cost routing, A/B testing models in production.
🔒 Enroll to unlock
Cost Engineering: Token Counting, Caching, Routing
Prompt caching (Anthropic up to 90% savings), model routing (Haiku for simple, Opus for hard).
🔒 Enroll to unlock
Project: Build a Customer Support Chatbot for Indian SaaS
Integrate Claude + Gemini with fallback, log conversations to PostgreSQL, track CSAT.
🔒 Enroll to unlock
Unit 8: RAG, Vector Databases & Prompt Engineering 7 topics
RAG Fundamentals + When NOT to Use It
Retrieval-augmented generation pipeline. When fine-tuning wins over RAG, and vice versa.
🔒 Enroll to unlock
Chunking Strategies & Document Processing
Fixed-size, semantic, sentence-window, parent-document retrieval. Handling PDFs, DOCX, HTML.
🔒 Enroll to unlock
Vector Databases: Pinecone, Weaviate, Qdrant, pgvector
Comparison, self-hosted vs cloud, filtering + hybrid search with BM25.
🔒 Enroll to unlock
Advanced RAG: Rerankers, HyDE, Multi-Query, Contextual Retrieval
Cohere/Voyage rerankers, Anthropic's contextual retrieval technique (49% error reduction).
🔒 Enroll to unlock
Evaluation of RAG Systems
Retrieval metrics (MRR, nDCG), generation metrics (Ragas, DeepEval, LLM-as-judge).
🔒 Enroll to unlock
Prompt Engineering Patterns
Few-shot, chain-of-thought, tree-of-thoughts, self-consistency. XML structure for Claude, role prompts for GPT.
🔒 Enroll to unlock
Project: RAG Over Indian Legal Judgments
Build a search + summarization system over Supreme Court + HC rulings; Qdrant + Claude.
🔒 Enroll to unlock
Unit 9: Agentic AI Systems (Single Agent & Multi-Agent) 8 topics
What is an Agent? (Signal vs Noise in 2026)
Tool use + planning + memory. When an LLM call isn't an agent; when a workflow isn't an agent.
🔒 Enroll to unlock
Single-Agent Systems with Tool Use
Anthropic tool-use API, OpenAI function calling. Building a SQL-querying agent.
🔒 Enroll to unlock
ReAct + Reflection Patterns
Reasoning + Acting loop, self-critique for quality improvement. When to reflect, when to not.
🔒 Enroll to unlock
Multi-Agent Systems: Orchestrator-Worker Pattern
Planner agent dispatches to specialist workers. Anthropic's multi-agent research architecture.
🔒 Enroll to unlock
LangGraph, CrewAI, AutoGen Comparison
Which framework for which use case. Graph-based state (LangGraph) vs role-based (CrewAI).
🔒 Enroll to unlock
MCP (Model Context Protocol) Integration
Claude's MCP for tools and data sources — connecting to Google Drive, Slack, GitHub, custom APIs.
🔒 Enroll to unlock
Evals & Safety for Production Agents
Sandboxing, prompt injection defense, output validation, audit logging, cost caps.
🔒 Enroll to unlock
Project: Resume-to-Interview-Prep Multi-Agent System
Parse resume → research company → generate mock questions → evaluate answers. 4-agent system with Claude.
🔒 Enroll to unlock
Unit 10: MLOps, Deployment & Capstone Placement Project 7 topics
Model Deployment: FastAPI + Docker
Wrap sklearn/PyTorch model in REST API. Dockerize. Deploy to Hostinger VPS / AWS EC2.
🔒 Enroll to unlock
Serving LLMs: vLLM, TGI, Ollama
Self-hosting open-source models. GPU memory math. Quantization (GGUF, AWQ, GPTQ).
🔒 Enroll to unlock
MLflow + Weights & Biases for Experiment Tracking
Runs, artifacts, model registry. W&B sweeps for hyperparameter tuning.
🔒 Enroll to unlock
CI/CD for ML: GitHub Actions + DVC
Data versioning, automated retraining pipelines, model tests (input schema, output range, latency).
🔒 Enroll to unlock
Monitoring: Drift Detection + LLM Observability
Data drift, concept drift. Langfuse/Helicone for LLM tracing, latency, cost, and quality monitoring.
🔒 Enroll to unlock
Resume & LinkedIn for 2026 Data Science Roles
How to frame projects, GitHub portfolio structure, Kaggle medals, LinkedIn SEO for recruiters.
🔒 Enroll to unlock
Capstone Project + Interview Rounds
End-to-end project: problem framing → data → model → deployment → report. Mock interview with ML/stats/SQL/case rounds.
🔒 Enroll to unlock
⭐ Student Reviews
No reviews yet. Be the first!
₹35,000
Enroll — ₹35,000
✨ 1-year access · Free updates included
30-day money back guarantee
Duration
120 Days
Format
Both Classes
Trainer
Expert Trainer
Students
0 enrolled
Language
Telugu / English
Certificate
On completion
Placement
80% rate