Senior Technical Lead

We are seeking a highly skilled Senior AI Engineer to lead the development of our next-generation Agentic AI systems. In this role, you will design, build, and scale advanced AI pipelines, focusing on complex RAG techniques, hybrid retrieval, and multi-agent workflows. You will work across the full stack-from engineering robust backend APIs to deploying scalable, secure AI solutions using Kubernetes and Docker.

You will play a critical role in bridging enterprise data with AI capabilities using the Model Context Protocol (MCP), while ensuring our systems are highly performant, secure, strictly governed, and free from hallucinations., * High-Performance Systems: Build highly scalable systems utilizing async and batch-parallel processing, caching mechanisms, and tuned vector search/database queries for massive document volumes.

• Traffic Management: Apply rate-limiting techniques including token bucket, sliding window, per-user/per-API quotas, and backoff/retry mechanisms for LLM and API calls.
• API Development: Develop robust, asynchronous REST APIs and webhooks (HMAC-SHA256 signed) using FastAPI.

Security, Governance & Observability

• AI Security: Protect against prompt injections, detect and mask PII, filter sensitive data from prompts and responses, and implement strict AI guardrails.
• Governance: Build robust AI governance frameworks utilizing SSO (Keycloak/OAuth2/OIDC), role-based access control (RBAC), audit logging, and systematic evaluation of model outputs.
• Monitoring & Observability: Monitor systems by tracking LLM latency, token usage, cost, response quality, and data drift using LangSmith, complete with alerts and observability dashboards.

• Python 3.11
• JavaScript / TypeScript
• Bash

Backend & APIs

• FastAPI, Uvicorn, Pydantic v2 / pydantic-settings
• httpx (HTTP/2), aiofiles, async I/O
• REST APIs and webhooks (HMAC-SHA256 signed)

LLMs & AI Frameworks

• Models: Diverse knowledge of LLMs including gpt-oss-120b (text), pixtral-12b (vision), multilingual-e5-large (embeddings) or similar equivalents.
• Frameworks: LangChain, LangGraph (agent orchestration with checkpointing), LangSmith (tracing).

RAG, Vector Search & Data Stores

• Milvus (pymilvus, langchain-milvus) or similar vector stores for vector storage and metadata-filtered search.
• LangChain text splitters for advanced chunking.
• MongoDB or similar NoSQL document stores.

Model Context Protocol (MCP)

• FastMCP and the MCP Python SDK.
• Building JWT-secured MCP servers that allow external AI assistants to utilize internal search tools.

Frontend

• React

Security & SSO

• Keycloak SSO (OIDC / OAuth2).
• Advanced PII masking, sensitive-data filtering, and application guardrails.

DevOps, Deployment & Infrastructure

• Docker, Kubernetes.
• Nginx (reverse proxy, SSL/TLS).
• RHEL 9 VMs, Git, and shell-script deployments.

Experience managing CI/CD and deployments across five environments (dev/test/stage/pre-prod/prod) using Methods of Procedure (MOPs) for releases.

Must Have Skill

Database -NoSQL - MongoDB,Python,Red Hat OpenShift,Retrieval-Augmented Generation (RAG),Vector Databases, * Proven experience building and deploying production-grade AI agents, LLM pipelines, and RAG systems.

• Deep understanding of asynchronous Python programming and scalable backend architectures.
• Hands-on experience with vector databases and semantic search techniques.
• A security-first mindset with experience implementing SSO, data masking, and LLM guardrails.
• Strong DevOps fundamentals with the ability to manage deployments across strict, multi-stage enterprise environments.

• Pipeline Design: Design and run automated pipelines for RAG and AI Agents, including advanced hybrid RAG (vector + keyword retrieval).
• Agentic Workflows: Develop multi-step agents built on LangGraph for complex tasks like document classification and multi-step reasoning. Develop the full stack and integrations for Agentic AI systems.
• Hallucination Control & Grounding: Ensure outputs are grounded in context through source-cited responses, retrieval relevance checks, and checking model outputs against trusted enterprise data.
• Multi-Modal Processing: Process text, images, and documents (PDF/DOCX) utilizing advanced vision, text, and embedding models.
• Prompt Engineering: Implement robust prompt engineering and systematic prompt management.