Manufacturing Company

Job Description

1. Profile:

Designation: Lead Data Science and AI

Grade: Senior Manager

Location: Head Office, New Delhi

Reports to: Head Digital, Data & AI

Team: Data Science Engineers (direct reports)

Qualification: B.Tech/M.Tech/MCA (Computer Science, Data Science or related)

Experience: 10-12 Years

2. Major Responsibilities:

Position Overview: We are seeking a hands-on Senior Data Science & AI professional who will design, build, and operate the organisation's full-spectrum, in-house AI capability covering:

- Data Science (Python / ML): classical machine learning models for business use cases

- Machine Learning Engineering: MLOps, deployment & monitoring of models in production

- AI Engineering Gen AI & Agentic AI: LLM applications, RAG solutions and autonomous agents

This is a player-coach role. The individual will personally build models, LLM/RAG applications and agents while leading a small team of Data Science Engineers. The mandate is to industrialise AI through MLOps/LLMOps so use cases move from pilot to sustained production value, using Python and modern ML/Gen AI stacks on the enterprise data & AI platform.

At our starting scale, this role also builds the internal AI platform (LLM gateway, vector stores, guardrails, evaluation tooling) and owns it jointly with the Lead Data and Analytics Engineer. The data & AI platform is envisioned on open-source technologies. The role demands genuine innovation appetite: exploring, building and maintaining new technologies internally rather than defaulting to packaged products.

Key Responsibilities:

1. Data Science & Machine Learning (Hands-on):

- Personally design, build and validate ML models in Python (scikit-learn, XGBoost, PyTorch/TensorFlow) for business use cases such as:

1. Demand forecasting, pricing and market/sales analytics

2. Manufacturing use cases: quality prediction, yield, energy optimisation, predictive maintenance

3. Supply chain, inventory and procurement optimisation

- Own the full model lifecycle: problem framing, feature engineering, training, validation, and business sign-off of model results.

- Work with AI-ready curated datasets on the enterprise data platform in collaboration with the Data & Analytics Engineering team.

2. Machine Learning Engineering MLOps, Deployment & Monitoring (Hands-on):

- Industrialise models through MLOps practices: experiment tracking, model versioning, CI/CD pipelines, containerised deployment.

- Deploy models as APIs/batch jobs integrated with business applications and operations digitalization use cases.

- Implement monitoring for drift, accuracy and data quality, with automated retraining pipelines and alerting.

- Ensure models remain healthy in production and deliver sustained value, not one-time pilots.

3. AI Engineering Generative AI & Agentic AI (Hands-on):

- Design and build LLM-based applications: knowledge assistants, document intelligence, content/code generation and Copilot-style experiences.

- Develop RAG solutions: chunking and embedding strategies, vector search, grounding, and retrieval quality tuning over enterprise knowledge.

- Build Agentic AI systems: multi-step autonomous agents and copilots that execute business workflows using frameworks such as LangChain/LangGraph or equivalent.

- Apply LLMOps: prompt/agent versioning, evaluation harnesses, hallucination and safety monitoring, cost and latency optimisation.

- Build solutions on the internal AI platform services described below, hardening them with every use case delivered.

4. AI Platform Ownership Initial Build & Joint Operation (Hands-on):

- Personally build the internal AI platform and evolve it into shared enterprise infrastructure, jointly owned with the Lead Data and Analytics Engineer.

- Core platform components to build and operate:

1. LLM gateway / model access layer unified access to open-weight models (self-hosted via vLLM/Ollama or similar) and API-based models, with routing, caching and cost/usage controls

2. Vector stores & RAG services embedding pipelines and vector databases (e.g., pgvector, Qdrant, Milvus, Chroma) exposed as reusable enterprise knowledge services

3. Guardrails & safety layer input/output filtering, PII protection, policy enforcement and human-in-the-loop controls

4. Evaluation & observability LLM/agent evaluation harnesses, tracing and monitoring (e.g., MLflow, Langfuse or equivalent open-source tooling)

- Establish platform standards: APIs, templates and reusable components so every new AI use case builds on the platform instead of starting from scratch.

- Make and document build-vs-buy recommendations with the Head Digital, Data & AI, with a default bias toward open source and in-house capability.

5. Innovation & Open-Source Technology Leadership:

- The data & AI platform is envisioned on open-source technologies: actively evaluate, prototype and adopt open-source frameworks across the ML/Gen AI stack (e.g., PyTorch, MLflow, Airflow/Kubeflow, LangChain/LangGraph, vLLM, FastAPI, Grafana/Prometheus).

- Demonstrate willingness to explore and build new technologies internally: run structured proofs-of-concept on emerging models, agent frameworks and tooling, and convert successful ones into production components.

- Own the discipline of maintaining open-source technologies internally: version upgrades, security patching, performance tuning, internal documentation and runbooks, treating the platform as a long-lived internal product.

- Build internal reusable IP: libraries, templates, accelerators so capability compounds in-house.

- Stay current with the fast-moving AI ecosystem and bring pragmatic recommendations.

6. Team Leadership Data Science Engineers:

- Lead, mentor and grow a team of Data Science Engineers; set coding, experimentation and documentation standards.

- Review models, prompts, agents and code personally; own the technical quality of everything the team ships.

- Plan sprints and delivery against the prioritised AI use-case portfolio; balance delivery with capability building.

7. Business Collaboration & Value Delivery:

- Work directly with business users, plant teams and Business Analysts to frame problems and define success metrics.

- Translate plant and business requirements into AI solutions that drive measurable outcomes ( impact, efficiency, quality).

- Communicate model behaviour, limitations and results honestly to non-technical stakeholders.

8. Responsible AI, Governance & Best Practices:

- Apply responsible-AI practices: human-in-the-loop controls, guardrails, bias checks and model risk documentation.

- Ensure data privacy and security compliance in ML and Gen AI workloads per internal IT and data policies.

- Maintain reproducibility: versioned data, code, models, prompts and evaluation results.

Required Skills & Competencies:

Technical (Must-Have):

- Deep hands-on expertise in Python and the ML ecosystem (pandas, scikit-learn, XGBoost, PyTorch or TensorFlow).

- Proven MLOps experience: model deployment, CI/CD, containerisation (Docker/Kubernetes), monitoring and retraining in production.

- Hands-on Gen AI engineering: LLM APIs and open-weight models, prompt engineering, fine-tuning, RAG architectures, vector databases.

- Experience building agentic systems (LangChain/LangGraph, function calling, tool use, agent evaluation) and LLMOps practices.

- Experience self-hosting and operating open-source AI infrastructure model serving (vLLM/Ollama or similar), vector databases, MLflow/Airflow-class tooling, Docker/Kubernetes maintained internally, not only managed cloud services.

- Strong SQL and comfort working on modern data platforms open-source lakehouse stacks and/or Microsoft Fabric / Snowflake / Databricks.

- Track record of taking AI use cases from pilot to sustained production value.

Manufacturing & Sales Domain (Highly Preferred):

- Experience applying ML/AI in manufacturing or process industry (MES, SCADA, historian data).

- Experience with sales & marketing analytics and DMS/CRM data.

- Understanding of SAP-centric enterprise data landscapes.

Leadership & Soft Skills:

- Player-coach mindset: personally delivers while raising the team's standard.

- Strong problem-solving and experimentation discipline; judgement on AI value vs. hype.

- Innovation appetite: curiosity for emerging open-source technologies, with the rigour to harden and maintain them internally for the long term.

- Comfort working directly with business users and plant stakeholders; clear communication of technical trade-offs.

- Ownership of deliverables end-to-end with a bias for production-grade work.