A production retrieval layer combining agentic chunking, hybrid vector and graph search, reranking, and automated RAGAS evaluation.
LangChain LangGraph RAGAS Langfuse Azure AI Search Vector Databases Graph Databases Python
~ / whoami
Hi, I'm Kanishk Kashayap
Lead AI & Data Engineer at Carl Zeiss AG.
Building production Gen AI systems — agentic workflows, RAG, and the secure cloud platforms behind them.
LangChain LangGraph RAG MCP
Scroll to explore
Impact_
Lead AI & Data Engineer with 5+ years of experience designing, building, and operating production AI, Generative AI, data, and cloud systems. Combines AI architecture with hands-on implementation across agentic workflows, retrieval-augmented generation, model and tool routing, evaluation, observability, secure backend integration, and production improvement. Experienced in translating product and stakeholder needs into technical roadmaps, scalable architectures, measurable quality criteria, and maintainable releases.
94% retrieval accuracy
achieved on an internal benchmark by designing a hybrid RAG architecture that combined embeddings, keyword search, metadata filtering, and reranking.
50% fewer agent errors
reduced production agent errors using RAGAS and Langfuse evaluations for relevance, groundedness, tool-call correctness, and response quality.
60% faster integration
designed AI-MCP to standardize AI-to-tool and AI-to-data interactions across 15+ products, reducing retrieval and integration time.
40% lower API costs
reduced query latency by 25% through model benchmarking, intent-based routing, and asynchronous FastAPI services.
€850,000 revenue
built governed Databricks pipelines that acquired more than 30 internal customers.
Skills_
Technologies I work with regularly
// Gen AI & Agentic
LangChain LangGraph RAG MCP ReAct Multi-Agent Systems Prompt Engineering
// Retrieval & Search
Agentic Chunking Embeddings Vector Databases Graph Databases Hybrid Retrieval Semantic Search Azure AI Search Reranking
// Evaluation & Observability
Langfuse RAGAS Groundedness Testing Tool-Call Evaluation
// ML & Deep Learning
PyTorch TensorFlow Keras OpenCV MLflow
// Backend & Cloud
Python FastAPI Azure Docker CI/CD
// Data & Delivery
Databricks dbt SQL Terraform Git
Featured Projects_
A selection of things I've built
A protocol standardizing AI–tool interactions for scalable, secure contextual data access across 15+ products at Carl Zeiss AG.
Python Azure Entra ID Azure API Design
Enterprise ETL pipeline powering master data consistency across 7 business units and 15 countries at Carl Zeiss AG.
Databricks dbt Unity Catalog Python CI/CD