Data Scientist Roadmap 2026 — India
Data science in India in 2026 is no longer a "learn pandas + sklearn" job. The bar has moved. Hiring teams at Razorpay, Fractal, Tredence, LatentView, Mu Sigma, and the global capability centres in Hyderabad and Bangalore now expect candidates who can train, deploy AND monitor models. They expect MLOps hygiene, AWS or GCP fluency, and the judgement to know when XGBoost beats a deep network. The "trained one model in a notebook" candidate is no longer competitive.
Duration
12 months · self-paced
Difficulty
Intermediate
Starting salary
₹6–18 LPA
Time commitment
10–12 hours / week
What does a Data Scientist actually do?
A Data Scientist builds predictive models that influence business decisions. Day-to-day: pose the problem, get the data, engineer features, train models, evaluate against business metrics, deploy to production, monitor for drift. In Indian product companies in 2026, "deploy to production" is a baseline expectation — not a bonus. ML Engineers handle infrastructure; Data Scientists own the model lifecycle from problem to monitoring.
This 12-month roadmap is calibrated to that bar. It starts from class-12 math, builds linear algebra + probability fast, then layers classical ML, deep learning in PyTorch, and full MLOps on AWS. You finish with an end-to-end capstone (data ingestion → training → deployment → monitoring) that lives in your GitHub. Indian DS offers from this path in 2025 ranged ₹6 LPA at services-company entry to ₹18 LPA at product-company mid-senior. Median ₹9.5 LPA. With 2–3 years experience, ₹22–28 LPA is within reach at Indian product companies.
You can do this self-paced or condense to 4 months with a structured cohort. Self-paced has a real failure mode at the MLOps stage — most Indian self-taught data scientists never deploy a model to production, which is exactly the gap that hiring panels probe. Plan to ship a deployed model by Month 9 even if it is messy.
Month-by-month plan
The 5-stage path
- 01 · Month 1–2Salary by end of stage: Internship: ₹20–35K / month
Math + Python foundations
Skills to learn
- Linear algebra for ML
- Probability + Bayes
- Calculus intuition for backprop
- Python proficiency
Tools you'll touch
- Python
- numpy
- Jupyter
Projects to build
- Implement linear regression from scratch
- Bayesian A/B test calculator notebook
Jobs to target
- · Data Science Intern
- 02 · Month 3–5Salary by end of stage: ₹6–8 LPA at this stage
Classical ML
Skills to learn
- Linear/logistic regression
- Decision trees + GBMs
- XGBoost deep dive
- Feature engineering
- Model evaluation (ROC, PR, calibration)
Tools you'll touch
- scikit-learn
- XGBoost
- LightGBM
Projects to build
- Credit-risk scoring model (Kaggle dataset)
- Customer churn prediction with SHAP explanations
Jobs to target
- · Junior Data Scientist (services)
- · ML Analyst
- 03 · Month 6–8Salary by end of stage: ₹8–12 LPA at this stage
Deep learning with PyTorch
Skills to learn
- Tensors + autograd
- CNN architectures
- RNN + transformer fundamentals
- HuggingFace fine-tuning
Tools you'll touch
- PyTorch
- HuggingFace
- Weights & Biases
Projects to build
- Image classifier on a domain-specific dataset
- Fine-tune a HuggingFace model for Indian-language text classification
Jobs to target
- · Data Scientist
- · ML Engineer (junior)
- 04 · Month 9–10Salary by end of stage: ₹10–18 LPA at this stage
MLOps on AWS
Skills to learn
- Git + Docker for ML
- MLflow tracking
- SageMaker training + endpoints
- CI/CD with GitHub Actions
- Drift monitoring
Tools you'll touch
- AWS SageMaker
- MLflow
- GitHub Actions
- Docker
Projects to build
- Deploy a churn model to SageMaker with autoscaling
- CI/CD pipeline for model retraining on new data
Jobs to target
- · Data Scientist at Razorpay, Fractal, Tredence, LatentView
- 05 · Month 11–12Salary by end of stage: ₹6–18 LPA first offer · median ₹9.5 LPA
Capstone + interview prep
Skills to learn
- ML system design — 4 reference problems
- Live coding mocks (LeetCode-style for DS)
- Resume + LinkedIn rewrite
Tools you'll touch
- Streamlit (for capstone demo)
- Github
Projects to build
- End-to-end capstone with Streamlit demo + writeup
- 3 system-design walkthroughs in your portfolio
Jobs to target
- · Data Scientist at Indian product companies + GCCs
The exact stack — and why each one matters
Python (numpy, pandas)
Daily driver
scikit-learn + XGBoost
Classical ML — still the #1 interview topic in Indian DS rounds
PyTorch
Deep learning framework Indian product companies standardise on
AWS SageMaker
Train + deploy at scale
MLflow
Experiment tracking + model registry
HuggingFace
Transformers + fine-tuning
Build these. Recruiters open them.
- 01Credit-risk scoring model deployed as a SageMaker endpoint
- 02Customer churn prediction with SHAP-explained dashboard
- 03Demand forecasting using time-series transformers
- 04NLP fine-tune for Indian-language customer support
- 05End-to-end capstone with monitoring + retraining pipeline
Where this path leads
- Year 1–2: Data Scientist · ₹6–18 LPA
- Year 3–5: Senior DS / ML Engineer · ₹22–35 LPA
- Year 5–8: Lead DS / Staff ML · ₹40–60 LPA
- Year 8+: Head of DS / Principal · ₹70 LPA – ₹1.5 Cr+
Five things people do wrong on this path
- 1Training models in notebooks and never deploying — this is the #1 thing hiring panels probe
- 2Going deep on deep learning before nailing classical ML — most Indian DS interviews still hinge on XGBoost
- 3Skipping the math primer — you do not need to derive backprop, but you need to understand it
- 4Ignoring AWS / cloud — production ML lives in the cloud
- 5Building a "Kaggle champion" portfolio with no production-style projects
Compress this into a 3-month cohort
Self-paced is free. A structured cohort with weekly mentor reviews + 50-partner placement support compresses the timeline and removes the common failure modes. Same content, faster outcome.
- Live cohort, max 15 students
- Weekly mentor reviews + project feedback
- 90-day placement support · 50+ hiring partners
- 3-month no-cost EMI · 7-day refund