Senior GenAI & Agentic AI Engineer

LLM Engineer | Agentic AI Engineer | AI/ML Engineer | NLP Engineer | MLOps Engineer

Organisation: TVS SCS (TVSSCS) | Location:Chennai,Madurai,Coimbatore | Experience: 3-7 Years | Type: Full-Time

Skills & Technology Keywords

This section is structured to support automated candidate matching on job portals and AI-powered ATS platforms.

AI / LLM Platforms & APIs

OpenAI GPT-4 / GPT-4o | Anthropic Claude | Google Gemini | Meta LLaMA 2 / LLaMA 3 | Mistral | Falcon | Cohere | Azure OpenAI Service | Google Vertex AI | AWS Bedrock | Hugging Face Hub | Ollama

Agentic AI & Orchestration Frameworks

LangChain | LlamaIndex | AutoGen | CrewAI | Semantic Kernel | AgentExecutor | ReAct Agents | Chain-of-Thought Prompting | Tool Use / Function Calling | Multi-Agent Systems | Autonomous Agents | Workflow Orchestration | Human-in-the-Loop (HITL) | LangGraph

RAG & Knowledge Retrieval

Retrieval-Augmented Generation (RAG) | Vector Search | Semantic Search | Hybrid Search | Re-ranking | Contextual Chunking | Document Embeddings | sentence-transformers | OpenAI Embeddings | BGE Embeddings | E5 Embeddings | FAISS | Pinecone | Weaviate | Milvus | Chroma | Qdrant | pgvector | Elasticsearch

ML / Deep Learning & Fine-Tuning

PyTorch | TensorFlow | Hugging Face Transformers | PEFT | LoRA | QLoRA | Supervised Fine-Tuning (SFT) | RLHF | DPO | Instruction Tuning | Model Quantization | ONNX | TensorRT | vLLM | TensorRT-LLM | Triton Inference Server | Distillation

MLOps & Model Lifecycle

MLflow | Weights & Biases (W&B;) | DVC | Kubeflow | Airflow | Prefect | Model Registry | Model Versioning | CI/CD for ML | GitHub Actions | GitLab CI | Jenkins | ArgoCD | Experiment Tracking | Data Versioning

Cloud & Infrastructure

AWS (SageMaker, EC2, S3, Lambda, ECS) | Azure (OpenAI, ML Studio, AKS) | GCP (Vertex AI, Cloud Run, GKE) | Docker | Kubernetes | Helm | Terraform | Serverless | GPU Infrastructure | NVIDIA CUDA | A100 / H100 GPUs

Backend & API Engineering

Python | FastAPI | Flask | Django | Node.js | REST API | GraphQL | Async Programming | Microservices | gRPC | WebSockets | API Gateway | OAuth2 | JWT | Rate Limiting

Databases & Storage

PostgreSQL | MySQL | Redis | MongoDB | Elasticsearch | pgvector | SQL | NoSQL | Schema Design | Query Optimization | Object Storage (S3/Blob)

Observability & Monitoring

Prometheus | Grafana | LangSmith | LangFuse | Arize AI | Evidently AI | OpenTelemetry | Structured Logging | Distributed Tracing | Cost Monitoring | Hallucination Detection | Token Usage Tracking | Agent Trace Logging

Prompt Engineering & AI Safety

Prompt Engineering | System Prompts | Few-Shot Learning | Zero-Shot Prompting | Context Window Management | Guardrails | Responsible AI | AI Governance | Bias Detection | Output Validation | Grounding | Factuality

About the Role

TVSSCS is hiring a Senior GenAI & Agentic AI Engineer who has shipped production AI systems to real users - not just built prototypes. You will own the architecture, deployment, and reliability of LLM-powered and agentic AI systems that run in production at scale within the TVSSCS engineering organisation. This is an engineering-first role; the expectation is production-grade code, observable systems,



and measurable business impact.

Key Responsibilities

1. Agentic AI System Design & Deployment

· Architect and deploy multi-agent orchestration pipelines using LangChain, LlamaIndex, AutoGen, CrewAI, LangGraph, or Semantic Kernel

· Build autonomous AI agents with tool-use, self-correction, ReAct loops, and long-horizon task execution

· Implement short-term, long-term, and episodic memory layers for stateful, context-aware agent behavior

· Design human-in-the-loop (HITL) workflows for high-stakes decision points within agent pipelines

· Integrate function calling and structured output enforcement with GPT-4, Claude, Gemini, and open-source LLMs

2. LLM Integration, RAG & Knowledge Systems

· Build and optimize production RAG pipelines using Pinecone, Weaviate, Milvus, Chroma, Qdrant, or pgvector

· Apply advanced retrieval: hybrid search (dense + sparse), cross-encoder re-ranking, contextual chunking, HyDE

· Manage embedding models (OpenAI, BGE, E5, sentence-transformers) and vector index lifecycle in production

· Implement multi-LLM routing (OpenAI, Anthropic Claude, Azure OpenAI, AWS Bedrock, Google Vertex AI) with fallback and cost controls

· Apply prompt engineering best practices: chain-of-thought, few-shot, system prompt design, context window management

3. Production ML Engineering & MLOps

· Own model deployment end-to-end: Docker, Kubernetes, CI/CD (GitHub Actions / GitLab CI / ArgoCD), rollback strategies

· Optimize LLM inference for low-latency, high-throughput serving using vLLM, TensorRT-LLM, Triton Inference Server, ONNX

· Manage ML lifecycle with MLflow, Weights & Biases, DVC - experiment tracking, model registry, versioning

· Execute efficient fine-tuning with LoRA, QLoRA (PEFT), SFT, and RLHF / DPO on proprietary datasets

· Deploy on AWS (SageMaker, ECS, Lambda), Azure (ML Studio, AKS), or GCP (Vertex AI, Cloud Run)

4. Observability, Monitoring & Reliability

· Build real-time GenAI monitoring dashboards using LangSmith, LangFuse, Arize AI, Evidently AI, or Prometheus + Grafana

· Track production GenAI metrics: token costs, hallucination rates, latency p50/p95/p99, agent tool-call traces

· Implement OpenTelemetry-based distributed tracing and structured logging for multi-agent pipelines

· Design alerting, circuit breakers, and graceful degradation patterns for AI service failures

· Conduct root-cause analysis on production incidents and drive SLA/SLO improvements

5. Backend Engineering & API Development

· Build scalable AI microservices with FastAPI, Flask, or Django; design REST and WebSocket APIs for real-time AI features

· Implement authentication (OAuth2, JWT), rate limiting, input validation, and structured error handling

· Integrate with PostgreSQL, Redis, MongoDB, Elasticsearch for data persistence and caching layers





· Apply async programming patterns (asyncio, Celery) for high-concurrency AI workloads

Must-Have Requirements

· 3-7 years of hands-on experience in ML Engineering, AI Engineering, or LLM Engineering roles

· MANDATORY: At least 2 production-deployed AI/ML systems (describe in application: problem, scale, your ownership, tech stack used)

· Proven experience deploying LLM applications - RAG pipelines, AI chatbots, autonomous agents, or AI-powered workflows - to production cloud (AWS / Azure / GCP)

· Hands-on experience with at least one agentic AI framework: LangChain, LlamaIndex, AutoGen, CrewAI, or Semantic Kernel

· Strong Python proficiency; experience with FastAPI or Flask for building AI-serving microservices

· Experience integrating at least one LLM API in production: OpenAI, Anthropic, Azure OpenAI, Cohere, or Hugging Face

· Hands-on experience with vector databases (Pinecone / Weaviate / Milvus / Chroma / pgvector / FAISS) in production RAG systems

· Solid command of Docker and Kubernetes for containerized AI model deployment

· Understanding of prompt engineering, context window management, and LLM cost optimization

· Exposure to MLOps tooling: MLflow, Weights & Biases, or DVC for experiment tracking and model management

Good to Have

· Experience with multi-agent frameworks: AutoGen, CrewAI, LangGraph, or custom orchestrators

· LLM fine-tuning using LoRA, QLoRA, PEFT, SFT, RLHF, or DPO on domain-specific datasets

· Inference optimization experience: vLLM, TensorRT-LLM, ONNX, model quantization (INT4/INT8), or distillation

· GenAI observability experience with LangSmith, LangFuse, Arize AI, or Evidently AI

· Knowledge of responsible AI, AI governance, output guardrails, bias detection, and factuality validation

· Open-source contributions to GenAI, LLM, or agentic AI tooling

· Frontend exposure (React, TypeScript) for building internal AI dashboards or chat interfaces

· Familiarity with enterprise AI compliance, data privacy, and security best practices (SOC 2, GDPR)

What We Expect From Applicants

Candidates must demonstrate real production experience. In your application, include:

· Production system #1 - problem solved, users/requests at scale, your specific role, key technologies used

· Production system #2 - same format as above

· One production incident you debugged and resolved - what broke, how you diagnosed it, what you fixed

· Links to GitHub, portfolio, blog posts, or open-source contributions (preferred but not mandatory)

Note : Applications without documented production AI project experience will not be shortlisted.

Why Join TVSSCS

· Build enterprise-grade AI systems used in real business operations - not demos or internal tools

· Work with state-of-the-art LLM infrastructure: GPT-4, Claude, Gemini, LLaMA, custom fine-tuned models

· High ownership: you own architecture decisions, not just feature tickets

· Learning & certification budget for AI/ML conferences, cloud certifications, and research

· Collaborative engineering culture with solid peer review, design discussions, and knowledge sharing

- · Growth path into AI Architecture / Principal Engineer roles within TVSSCS latforms or

📌 GEN AI Engineers (Madurai)
🏢 TVS Supply Chain Solutions
📍 Madurai