Senior Staff Data Scientist · G13

Central Data Science (CDS) · Bengaluru · Flipkart

About the Team

Central Data Science (CDS) is the 120-person team that builds the intelligence layer underneath Flipkart — Search, Recommendations, Pricing, Ads, Planning, Supply Chain, Fintech, Trust & Safety, Catalog, and the AI platforms they all run on. We are structured as domain Pods, each owning a problem space end-to-end alongside its product and engineering counterparts. Furthermore we are responsible for building AI platforms which are adopted across Flipkart group.

What makes the environment unusual is the combination of three things you rarely get together: scale that actually stress-tests models (BBD runs Flipkart's systems at 6–7× normal traffic, with the data layer crossing a million QPS at peak); consequential surfaces — Search, Pricing, Forecasting, Fraud, Reco, Ads — that decide what hundreds of millions of shoppers see, pay, and receive; and an AI-native pivot already in production, not on a roadmap deck — agentic Search (AI Mode), Agent Copilot in CX, AutoQC across catalog, etc. If you want to ship into this transition, not narrate it, this is the team.

About the Role

Senior Staff Data Scientist is the domain ownership rung on the CDS IC path.. The scope is a Pod, or a tightly related set of surfaces within a business unit (domain Pod). You own the DS technical agenda for that domain — methodology, evaluation, production rollout, impact measurement — and you are the highest-density DS presence your leadership across Pod treats as the technical authority. Much of the work starts ill-defined: a business problem stated loosely, and your job is to turn it into the right DS formulation before anyone hands you a metric.

The clearest line we draw: a Senior DS is given a problem and trusted to solve it well; a Senior Staff DS is trusted to decide which problems the Pod should be solving in the first place and drive adoption and impact. Above this level, scope widens to multi-domain strategy and org-level architecture.

What You Will Do

Own a domain metric end-to-end. Take a primary metric such as CM, GMV, or relevance lever in your Pod and own it across modelling, experimentation, and production rollout.
Drive the methodology choices that shape what the Pod can measure. Decide on baselines, evaluation design, and validation discipline. Make the cost/latency/sophistication tradeoffs and defend them in plain language to your PM and EM.
Own the DS technical roadmap for your domain. Decide what the Pod should work on over the next 2–4 quarters — which problems, which methodology bets, in what order. Earn the right to set direction; defend it when challenged.
Be the DS technical counterpart to PM and EM. Surface data-backed constraints early. Align on data contracts, feature definitions, and labelling standards. Translate model behaviour into decisions, not metric dumps.
Raise the bar. Primary technical mentor for G10–G12 DSs in your Pod. Set the standard for experimentation rigour, code, and written communication.

What You Will Need

Mandatory

B.Tech / M.Tech in CS, Statistics, Mathematics, or a closely related quantitative field — or equivalent capability demonstrated through deployed systems, publications, or a verifiable technical portfolio. PhD is valued, not required.
11+ years in applied ML/DS, with at least 4–5 in a role with real technical ownership.
A track record of shipping ML/DS systems with measurable product or business impact at meaningful scale — work you can speak to as an owner, with outcomes you can quantify.
Strong command of the statistical and ML fundamentals relevant to your target domain. You should be able to defend your methodology choices under pressure and name the failure modes of your own systems before someone else does.
Production-grade Python; fluency with the standard ML/DS toolchain (sklearn, PyTorch or TensorFlow, experiment tracking, feature stores).
Demonstrated experience working directly with product and engineering inside a cross-functional Pod or squad model.
Ability to make complex technical tradeoffs land with a non-specialist audience.

Domain Depth (matched to the hiring Pod)

Candidates are evaluated against the specific Pod they are interviewing for. Active areas: Search & Retrieval · Recommendations · Pricing & Promotions · Supply Chain & Planning · Ads & Monetisation · Retail AI (catalog, visual, multimodal) · Fintech & CX · Trust & Safety · AI Platforms.

Strongly valued, not mandatory

Built from scratch or re-architected a complex, large-scale ML/DS system — Search, Ads, Pricing, Trust & Safety, or similar — including the production excellence that keeps it healthy (monitoring, drift detection, rollback) on infrastructure like Spark/Hive and distributed feature computation.
GenAI application development — RAG, agent design, and LLM evaluation discipline (golden sets, LLM-as-judge tradeoffs, drift definition, rollback automation).
Modular DS components others build on — feature pipelines, evaluation harnesses, reusable baselines.
Deep experience in a high-growth e-commerce, marketplace, or consumer-internet environment.
A public technical presence — publications, conference talks, open-source, or technical writing that demonstrates depth.

How We Level

This posting is calibrated to Senior Staff. Leveling is part of the evaluation, not a barrier to applying — so a couple of honest notes:

If your experience is primarily notebook-to-deck analytics without production ML, this likely isn't the right fit at any IC level.
If you're earlier on the IC path — strong execution, but direction still largely set for you — you may be a better fit at our Staff or Senior DS levels. Tell us, and we'll evaluate you there. These are respected roles here, not lesser ones.
If you're already operating at multi-domain or org-level scope — strategy, architecture, executive influence — you may be a better fit at Principal. Call it out, and we'll level and evaluate accordingly.

Senior Staff Data Scientist I

Senior Staff Data Scientist