Backend AI Engineer

Job Title :Backend AI Engineer

Location : All over India

Notice period : Immediate joiner

Role Overview

• Job Title: Senior Backend AI Engineer (Agentic AI, RAG, LLM Engineering)
• Experience Required: 7+ Years
• Core Focus: Production-ready Agentic AI platforms, Retrieval-Augmented Generation (RAG), and full-lifecycle LLM engineering (data prep, fine-tuning, inference, and cloud deployment).

Key Responsibilities

• Agentic Platforms: Build backend platforms for autonomous workflows, orchestration, tool-calling, state/context management, and safety guardrails.
• RAG Systems: Design and optimize end-to-end RAG pipelines (chunking, embeddings, vector search, metadata filtering, reranking, and retrieval evaluation).
• LLM Fine-Tuning: Develop training pipelines, covering dataset curation, preprocessing, labeling strategies, hyperparameter tuning, and rollback mechanisms.
• Inference & Serving: Deploy high-availability, low-latency inference services with monitoring for hallucinations, retrieval precision/recall, and tool-call accuracy.
• Microservices & APIs: Build containerized AI APIs and asynchronous, event-driven microservices integrated with vector stores and model-serving stacks.
• Infrastructure & Kubernetes: Architect K8s deployment topologies, autoscaling policies (HPA), and disaster recovery plans for AI workloads.
• CI/CD & MLOps: Own end-to-end CI/CD and MLOps practices, including blue/green or canary releases, model versioning, prompt regression tests, and benchmarking.
• Collaboration & Mentorship: Bridge research prototypes to production alongside product and infra teams; mentor engineers via design reviews and coding standards.

Required Skills & Technologies

• Experience: 7+ years of backend engineering with proven production LLM/AI deployment experience.
• AI/LLM Core: Deep knowledge of Agentic AI frameworks, multi-step orchestration, and fine-tuning methods (PEFT, LoRA, QLoRA, instruction tuning).
• Vector Databases: Hands-on experience with FAISS, Pinecone, Weaviate, pgvector, or Milvus.
• Programming & APIs: Advanced Python, asynchronous programming, and API frameworks (FastAPI, Flask).
• Containers & Orchestration: Docker, Kubernetes (Helm, ingress, secrets, HPA, resource quotas).
• Datastores & Caching: PostgreSQL, Redis, object storage (S3/GCS), and NoSQL databases.
• DevOps & Cloud: CI/CD tools (GitHub Actions, GitLab CI, Jenkins), IaC basics, and core cloud infrastructure (AWS, GCP, or Azure).
• Observability: Prometheus, Grafana, OpenTelemetry, ELK/OpenSearch, and alerting pipelines.

Preferred Qualifications

• Serving Engines: Triton Inference Server, vLLM, TGI, TorchServe, or BentoML.
• ML Lifecycle Tools: MLflow, Weights & Biases, DVC, or equivalent experiment-tracking tools.
• GitOps: ArgoCD, Terraform, and declarative infrastructure workflows.
• AI Security: Knowledge of prompt-injection defense, data privacy, guardrail policies, and AI auditability.

Core Expectations

• Deliver scalable, highly reliable, and cost-efficient AI backend architectures.
• Effectively balance trade-offs between model accuracy, latency, and compute cost.
• Drive technical roadmap decisions and communicate clearly across cross-functional teams.