I'm Kartik Kumar - an AI Engineer delivering LLM-powered systems, multi-agent pipelines, and RAG architectures from prototype to production.
More informationExperienced in building multi-agent pipelines and RAG architectures in enterprise environments. Proficient in Python, LangChain, LangGraph, and cloud platforms including AWS and Azure. Demonstrated measurable impact on efficiency, cost reduction, and engineering quality across automotive and technology sectors.
Architected end-to-end orchestrated workflows across specialized agents connected to enterprise knowledge sources via MCP servers, driving 80% increase in user task efficiency through intent-driven agent routing and RAG at scale.
Python, Azure OpenAI, DataBricks, LangChain, LangGraph, CrewAI, MCP, FastAPIDelivered a GenAI-powered quality check for engineering requirements and test cases using Azure OpenAI and DataBricks, achieving 85%+ alignment with human evaluations. Built consistency check framework using prompt-tuned LLMs to automate validation of large-scale documentation.
Azure OpenAI, DataBricks, Prompt Engineering, LLM EvaluationStreamlined test case generation using LLMs - increasing coverage by 25% and cutting manual QA effort by 40%. Implemented Knowledge Graph-based RAG systems improving structured data retrieval accuracy across Mercedes, Jaguar, and Marelli.
Python, LangChain, GraphRAG, CrewAI, AWS, AzureAI-powered content automation that scrapes real-time trends, generates contextual posts with AI visuals, and publishes on schedule - fully autonomous end-to-end.
Python, CrewAI, LLMs, Web Scraping, Image GenerationUnified API with multi-signal routing engine that selects LLMs based on task complexity, confidence, risk, and latency - with real-time analytics on cost savings.
Python, FastAPI, SQLite, Streamlit, LLMsEnd-to-end cost management with FastAPI and Prometheus for real-time financial analytics, anomaly detection, and Dockerized deployment.
Python, FastAPI, SQL, Prometheus, DockerML pipeline identifying individuals from gait patterns using TensorFlow and Random Forest, achieving 20% accuracy improvement over prior methods.
TensorFlow, Keras, Random Forest, PyQt6Predictive model forecasting Billboard chart success for Spotify tracks, achieving 85%+ accuracy and 15% improvement in recommendation efficiency.
PyCaret, Extra Trees ClassifierReal-time traffic monitoring using YOLO and OpenCV for density analysis, reducing manual reporting errors by 25%.
OpenCV, YOLO, NumPy, PandasFacial recognition attendance with voice-enabled queries, cutting manual tracking time by 40%.
OpenCV, Face Recognition, Pandas, pyttsx3Have an idea or a challenge? Let's talk about how AI can solve it.