Data science in India has exploded as a career path over the last 5 years. Bootcamps, online courses, and self-taught entrants have created a deep — but uneven — talent pool. The challenge for foreign employers is filtering the production-capable practitioners from the notebook-only candidates. The pool that can deploy a model to production, monitor it, and iterate is much smaller than the headline count suggests.
Where the real data science talent sits
- Bengaluru — largest pool by far. Strong concentration of senior data scientists with 5+ years in production environments (Flipkart, Swiggy, Razorpay, PhonePe, Microsoft, Adobe alums).
- Hyderabad — second-largest pool. Microsoft, Amazon, Google, Salesforce, Qualcomm all have major data teams.
- Pune — smaller but high-quality, especially in pharma / healthcare data (focused on production systems, model validation, regulatory rigour).
- NCR (Gurgaon / Noida) — strong in consulting + fintech data science (American Express, Mastercard, ZS Associates).
- Mumbai — concentrated in financial-services data science (Morgan Stanley, JPMorgan, Macquarie).
2026 data scientist salary bands
Total annual CTC in INR by experience and city. Add 20-40% for ML research / applied scientist roles vs. standard data scientist.
| Level | Years exp | Bengaluru CTC | Tier-2 CTC |
|---|---|---|---|
| Junior Data Scientist | 0-2 | ₹10-22L | ₹8-16L |
| Data Scientist | 3-5 | ₹22-40L | ₹18-30L |
| Senior DS / Applied Scientist | 5-8 | ₹40-75L | ₹32-58L |
| Staff DS | 8-12 | ₹75-130L | ₹58-100L |
| Principal DS / ML Engineer | 12+ | ₹130-220L+ | ₹100-170L |
The skills test that filters
A typical 3-round technical evaluation that separates production-capable candidates from notebook-only ones:
- Take-home assignment (4-6 hours, do over 1 week) — given a real-world dataset, train a model, evaluate it, write a brief explaining trade-offs and what you'd do differently in production. Filters out candidates whose 'data science' is following tutorial steps.
- Live coding (60 min) — implement a basic ML algorithm from scratch (linear regression, decision tree, or k-means). Filters out candidates who only know how to call sklearn.
- System design (60 min) — design an end-to-end ML system for a specified problem (recommendation, fraud detection, demand forecasting). Discuss data pipelines, training infrastructure, deployment, monitoring, A/B testing. Filters out candidates who've never deployed anything.
Compensation calibrated to your specific role
FastLegal's hiring consultant returns a salary band calibrated to live offer data we see across 900+ clients — by level, city and specialty. For data scientists this matters because the range is wider than backend roles; ML research can be 1.4-1.8x of generalist DS at the same level.
Specialty premiums within data science
- Generalist DS — baseline (use the table above).
- MLOps / ML platform — 1.10-1.25x. Smaller pool, high demand.
- Applied ML for recommendations / ranking — 1.15-1.30x.
- ML research / applied scientist — 1.40-1.80x. Includes deep learning research, NLP / LLM specialists.
- Causal inference / experimentation — 1.10-1.20x.
- Computer vision specialists — 1.15-1.30x.
Common hiring traps for foreign employers
- Over-weighting Kaggle competition ranks — Kaggle skills don't always translate to production ML. Useful signal but not decisive.
- Hiring 'data scientist' for what is actually data engineering work — the production data pipeline + warehousing skill set is more scarce than predictive modelling.
- Pay scales too low — if your offer is at the bottom of the Bengaluru band, you'll lose to Indian SaaS unicorns that pay top-quartile.
- Ignoring infrastructure familiarity — a great modeller who has never deployed on AWS / GCP / Azure won't ship in your environment.
Frequently asked questions
Should we look for PhD candidates?+
For research roles, yes — IIT / IISc / IIIT PhD pool has serious depth. For applied roles, MS + work experience often beats PhD.
How long does a senior DS hire take?+
4-8 weeks from search to signed offer. Strong candidates have 3-5 active offers at any time; move fast.
Do we need to differentiate between Data Scientist and ML Engineer in our hiring?+
Yes — they're different roles. DS focuses on modelling, experimentation, statistical rigour. MLE focuses on production systems, scalability, latency. Hire to the actual job.
What about LLM / GenAI specialists?+
Very hot in 2026. Demand exceeds supply; expect to pay 1.5-2x of generalist senior DS. Most candidates with real LLM production experience have 2+ years of focused work in this area.
Stop reading circulars. Start running clean payroll.
Every FastLegal plan ships with a dedicated payroll consultant — a real human who runs your PF, ESI, PT, TDS and Form 16 issuance, configured to your salary structure, your state, and your hiring plan. You sign off. We do the rest.