Staff Data Scientist

About the Role As a Staff Data Scientist, you will architect and lead efficient solutions across key domains such as Recommendation, Search, Ads, and Discount Optimization. You will mentor cross-functional teams, review technical architectures and maintain a high standard for holistic solution design. Driving innovation, you will foster a culture of adopting (SOTA) models. Leveraging deep expertise in ML, DL and advancements like GenAI/LLMs, you will guide teams in acquiring new skills and integrating evolving paradigms into production systems. You will operate as a senior technical leader for the Food charter: shaping problem formulation, influencing product roadmaps, and ensuring our AI systems are robust, low-latency, and business-outcome driven What You’ll Do Own and drive the technical roadmap for AI systems across Search, Recommendations, Ads and Discounting, from problem framing to production rollout and post-launch iteration. Architect end-to-end ML/AI pipelines (data, models, orchestration, evaluation, monitoring) that meet strict constraints on latency, scale, cost and reliability. Evaluate and introduce SOTA techniques (retrieval/ranking, bandits/RL, causal uplift, GenAI/agents) in a pragmatic, production-ready manner. Partner with Product and Business to define success metrics, set up robust experimentation, and tie AI investments clearly to business KPIs and ROI. Provide architectural and design reviews for high-stakes ML systems across the Food charter; set and enforce engineering and MLOps best practices. Mentor and uplevel Senior/Lead Data Scientists and MLEs; act as a thought partner to leadership on build-vs-buy, platform strategy and multi-year bets. Represent Swiggy AI in internal and external forums (tech talks, blogs, publications, conferences) and help build the brand for Food AI. Skills & Experience Core Technical Skills Deep expertise in classical ML, representation learning and modern deep learning (e.g., transformers, two-tower/rec models, ranking architectures). Search & Recommendations : large-scale retrieval, LTR, multi-stage ranking, vector search, multi objective ranking and personalization.Fine-tuning SLMs/LLMs, building RAG/agentic workflows, conversational or copilot-style systems. Ads & Discounting : Bandits/RL for allocation, uplift/causal models, constrained optimization for budgets/pricing. Risk/Fraud: graph-based models (e.g., entity graphs, GNNs for fraud rings), classical ML (GBMs, tree ensembles, logistic regression), and deep learning frameworks such as PyTorch/TensorFlow, with hands-on work on encoder/transformer architectures (SLMs, BERT-style models) for representation learning, feature extraction and risk scoring in real-time systems Hands-on experience with: Strong system design skills for low-latency, high-throughput ML platforms; fluency with Python, PySpark, PyTorch/TensorFlow, feature stores, vector DBs and modern MLOps. Architectural & Strategic Skills Ability to design AI-first systems end-to-end: data contracts, feature pipelines, serving architecture, feedback loops and continuous learning. Comfort evaluating open-source vs proprietary models/tools, with clear reasoning on cost, risk, scalability and maintainability. Experience with orchestration / agent frameworks (e.g., LangGraph, CrewAI, AutoGen) and concepts like ontology layers, graph knowledge bases and multi-agent workflows. Strong product thinking: can connect model capabilities to customer journeys and business KPIs, and influence roadmaps accordingly. Leadership & Collaboration Demonstrated experience as a tech lead / staff-level IC, influencing multiple pods or domains. Ability to mentor and coach senior DS/MLEs, drive technical standards, and create a high-bar culture for experimentation and measurement. Excellent written and verbal communication; can simplify complex ideas for non-experts and build alignment across Product, Engineering and Business stakeholders. Bias for action, ownership and comfort working