Senior AI Engineer - Agentic Systems & RAG | Devsinc

In this role, you will own the end-to-end AI lifecycle—from initial experimentation, RAG system design, and prompt architecture to model evaluation, GPU inference optimization, containerized cloud deployment, and continuous MLOps monitoring. You will build high-throughput backend APIs, deploy autonomous AI agents with function-calling capabilities, and optimize retrieval pipelines over multi-modal vector stores. Working closely with enterprise clients, cross-functional engineering teams, and junior developers, you will guide technical decisions, mentor engineers, and ensure all AI systems are reliable, secure, cost-efficient, and aligned with measurable business outcomes.

Key Responsibilities

Generative AI Architecture, RAG & Agentic Workflows

• End-to-End AI System Design: Design, develop, and deploy AI/ML and LLM-based applications, including autonomous AI agents, tool-using systems, and human-in-the-loop decision workflows.
• RAG & Semantic Search Pipelines: Build, evaluate, and optimize Retrieval-Augmented Generation (RAG) and semantic search platforms using document chunking, embeddings, vector search, cross-encoder reranking, and grounding techniques.
• Backend API & Microservices Development: Construct robust, production-grade backend APIs and real-time inference microservices using Python (FastAPI, Flask, or Django) connected to PostgreSQL/SQL databases.
• Model Tuning & Optimization: Improve model inference quality, latency, throughput, and token costs through hyperparameter tuning, model quantization, dynamic batching, caching (Redis), and prompt optimization.

MLOps, Cloud Infrastructure & Governance

• Production MLOps & CI/CD: Implement automated testing, experiment tracking, model versioning, regression testing, and CI/CD deployment pipelines supported by benchmark evaluation datasets and human review loops.
• Cloud Architecture & Security: Guide cloud deployment across AWS, Azure, or GCP, enforcing security guardrails against prompt injection, data leakage, unauthorized access, and unsafe model outputs.
• Technical Mentorship & Client Leadership: Evaluate emerging AI technologies, mentor software engineers, document technical designs, and articulate complex architectural tradeoffs to clients and executive stakeholders.

Requirements

Technical Depth & Engineering Experience

• Educational Background: Bachelor's degree in Computer Science, Software Engineering, Data Science, or a related quantitative discipline.
• Professional Experience: 4+ years of post-graduation professional experience in AI/ML engineering, with a proven track record of shipping LLM or generative AI applications into production environments.
• Python & ML Frameworks: Production-level Python proficiency paired with hands-on experience in PyTorch and/or TensorFlow, alongside deep knowledge of neural networks, NLP, and feature engineering.
• LLM Integration & Tooling: Practical experience integrating commercial and open-source LLMs—including prompt engineering, structured JSON outputs, function calling, and context window management.
• Vector Search Expertise: Hands-on experience building semantic search pipelines using vector databases such as pgvector, Pinecone, Weaviate, Qdrant, Milvus, or Elasticsearch.
• API & Database Engineering: Proven experience building RESTful microservices with FastAPI, Flask, or Django backed by SQL databases (PostgreSQL, MySQL).
• Cloud, DevOps & Containers: Proficiency with Git, Docker, CI/CD pipelines, and cloud platforms (AWS, Azure, or GCP), combined with MLOps best practices for model tracking and monitoring.
• Evaluation & Security Foundations: Strong understanding of LLM evaluation frameworks, regression testing, human-in-the-loop validation, and AI security risks.

Preferred Qualifications

• Experience with orchestration frameworks (LangChain, LlamaIndex, LangGraph) and LLM evaluation tooling (LangSmith, Langfuse, Ragas).
• Hands-on experience with Hugging Face Transformers, vLLM, Ollama, or fine-tuning techniques (LoRA / PEFT).
• Familiarity with MLOps infrastructure tools (MLflow, Kubeflow, Kubernetes, Terraform) or multi-GPU acceleration setups.
• Experience with asynchronous task queues and messaging brokers (Airflow, Redis, Celery, Kafka, or RabbitMQ).
• Knowledge of advanced graph retrieval (GraphRAG), knowledge graphs, recommendation engines, computer vision, or A/B experimentation platforms.