Manager

Job Description: Manager Generative AI / Agentic AI & MLOps

Location : Gurugram - On Site

Reports To : SVP Data Science / AI

Role Purpose :

This role is responsible for designing, building, and deploying end-to-end Generative AI and Agentic AI systems, with a strong focus on local LLM deployment, secure enterprise GenAI applications, and scalable MLOps pipelines.

The position requires expertise in:

Local RAG (Retrieval-Augmented Generation) pipelines Agentic AI frameworks (multi-step reasoning systems)Private LLM hosting and optimization Robust MLOps architecture (Kafka, Kubernetes, Airflow, CI/CD)

Success in this role means delivering production-grade GenAI systems that are scalable, secure, cost-efficient, and deliver measurable business impact.

Key Result Areas & Responsibilities:

1. GenAI & Agentic AI Development (35%)

• Design and build end-to-end GenAI applications from scratch
• Develop Agentic AI systems using frameworks like:

LangChain, LlamaIndex, LangGraph

AutoGen, CrewAI, Semantic Kernel

• Implement multi-agent workflows for decision-making and automation
• Optimize prompt engineering, memory management, tool usage, and reasoning flows
• Outcome: Production-grade GenAI and Agentic AI applications delivering automation and intelligence

2. Local LLM Deployment & Optimization (20%)

• Deploy and manage local/private LLMs (on-prem or VPC environments)
• Work with models suchas:

LLaMA, Mistral, Mixtral, Falcon, Gemma,Qwen

• Use inference framewoks: vLLM, Ollama, Hugging Face Transformers, TensorRT-LLM
• Optimize for:

Latency, throughput, and cost

Quantization and model compression (GGUF, INT 4/8)

• Outcome: Secure, low-latency, cost-efficient LLM deployments

3. RAG Pipelines & Knowledge Systems (20%)

• Design and implement advanced RAG pipelines:

Document ingestion, chunking, embedding, retrieval, re-ranking, Grounding

• Build local knowledge bases using:

Vector DBs: FAISS, Chroma, Weaviate, Pinecone (optional hybrid), Qdrant

• Implement:

Hybrid search (BM25 + vector)

Context window optimization

• Develop domain-specific assistants and copilots
• Outcome: Accurate, grounded GenAI systems with high retrieval precision

4. MLOps, Data Engineering & Deployment (15%)

• Develop scalable MLOps pipelines using:

Kubernetes (container orchestration)

Kafka (real-time streaming)

Airflow (workflow orchestration)

• Implement CI/CD pipelines for ML:

GitHub Actions, Jenkins, GitLab CI

• Enable:

Model versioning, monitoring, and rollback

Logging, observability (Prometheus, Grafana)

• Outcome: Reliable, scalable, and automated AI deployment pipelines.

5. Governance, Security & Responsible AI (10%)

• Implement guardrails for GenAI:

Prompt injection protection

Output validation and filtering

• Ensure compliance with:

Data privacy policies

AI governance frameworks

• Manage model risk and audit readiness
• Outcome: Secure and compliant AI systems ready for enterprise deployment

Detailed Responsibilities :

• Planning

Define roadmap for GenAI, Agentic AI, and LLM adoption

• Identify use cases for automation, decisioning, and productivity gains
• Plan infrastructure for local AI deployment

Operational

• Build and deploy GenAI applications end-to-end
• Maintain pipelines for training, inference, and evaluation
• Ensure high uptime and system reliability

People

• Mentor team on GenAI frameworks and MLOps practices
• Drive capability building in LLMs, RAG, and Agentic systems

Governance

• Ensure ethical AI usage and safe deployment practices
• Maintain documentation for audit, compliance, and scalability

Improvement

• Continuously optimize :

Latency

Cost per inference

Model performance

• Build reusable AI components and frameworks

Technical Skills &Tools :

• GenAI & Agentic AI

LangChain, LlamaIndex,LangGraph

AutoGen, CrewAI, Semantic Kernel

Prompt engineering frameworks

• LLMs & Deployment

LLaMA, Mistral, Falcon, Gemma, Qwen

vLLM, Ollama, Hugging Face

Quantization tools (GGUF, bitsandbytes)

• RAG &Retrieval

FAISS, Chroma, Weaviate, Qdrant

BM25, hybrid retrieval techniques

Embedding models (BGE, OpenAI, Instructor)

• MLOps & Infrastructure

Kubernetes, Docker

Kafka (streaming pipelines)

Airflow (workflow orchestration)

CI/CD: GitHub Actions, Jenkins

• Programming & Data

Python, SQL, Spark

FastAPI,REST APIs

Distributed systems and microservices

• Monitoring & Observability

Prometheus, Grafana

MLflow, Weights & Biases

Qualifications & Experience :

Education - Bachelors/Masters in CS, AI, ML,or related

Experience -6-10 years Relevant

Experience - 5+ years in ML/AI, 2+ years in GenAI/LLMs

Industry - Fintech, Banking, AI-first tech companies

preferred Certifications - Cloud (AWS/GCP/Azure), ML, GenAI

Mandatory Skills: Generative ai, Agentic Ai, ML Pipelines, Retrieval Augmented Generation

Desirable Skills: Stakeholder Management, Cloud Computing