Skip to main content
Kaushalendra Singh (Marcus)

Kaushalendra Singh (Marcus)

AI Engineer | LangGraph, RAG & MCP Systems | Full Stack (Next.js, Node.js)

Talent
KanpurMember since August 31, 2026
TypeScriptSQLNext.jsReactZustandNode.jsPrismaKafkaRedisLangChainLangGraphOpenAIGeminiAnthropicMCPNeo4jPineconeMongoDBPostgreSQLDockerCI/CDGroqOllamaVS Code APISocket.ioPWAElectronTwilioZodPythonAwsFAISSFastAPIPrompt EngineeringRAGLangSmithLangfuseLLM Evals
·

Hire this person

Send a job offer directly to this candidate

TypeScriptSQLNext.jsReactZustand

About

I am an AI Engineer focused on building agentic systems that reason, retrieve, and act, not just chat. I design multi-step agent pipelines with LangGraph, build RAG systems that combine vector and graph retrieval, and integrate LLMs into production tools through MCP servers and structured tool-calling. My background also includes full stack development with Next.js, Node.js, and PostgreSQL, so I can take an AI system from a working prototype to a shipped product end to end.

I am currently working as an AI Founding Engineer Intern at EpicWise, where I build the core context engine and public MCP server used by the platform. Open to AI Engineer, Software Engineer, and Full Stack roles where I can build systems that make AI genuinely useful for real users.

Experience

AI Founding Engineer Intern, EpicWise

May 2026 to Present

Improved the accuracy and reliability of the platform's reasoning by designing multi-step retrieval and reasoning pipelines using LangGraph, working directly with the founding team on the core context engine.

Enabled external developers and agents to safely access the platform by building EpicWise's public MCP server, including authentication, authorization, an agent harness testing agent, and a requirement-analysis agent.

Increased answer accuracy for complex user queries by building a hybrid retrieval system combining vector search with Pinecone and graph-based retrieval with Neo4j, along with the system prompts, execution routers, and tool-calling schemas that route queries between them.

Full Stack Developer Intern, GRS Worker

Sep 2025 to Nov 2025

Supported reliable scaling of the platform by building a backend with Node.js, Prisma, and PostgreSQL, using Redis caching to keep response times fast under load.

Reduced data errors by 40% by introducing Zod validation and structured service-layer workflows across the application.

Made deployments faster and more consistent by integrating Twilio and QR-based systems and containerizing all services with Docker and CI/CD pipelines.

Education

Bachelor of Technology – Harcourt Butler Technical University, Kanpur

Skills

AI EngineerLangGraphRAG & MCP SystemsFull Stack (Next.jsNode.js)

Reviews

Related