Rakesh Kumar

About Rakesh Kumar

AI Engineer | LLM & RAG Architect | MLOps SpecialistBuilding production-ready AI systems that transform enterprise operations and unlock intelligent automationI am GenAI and MLOps specialist with over 9 years of experience designing, deploying, and scaling advanced machine learning and deep learning systems. My work spans from foundational ML models and big data pipelines to cutting-edge LLM architectures and autonomous AI agents—all with a focus on real-world impact, scalability, and operational efficiency. Philosophy:I believe great products aren\'t just built—they\'re researched. With intentional inquiry before the first line of code, we can architect AI systems that aren’t just smart, but truly reliable, interpretable, and impactful. Research is the bridge between a good product and a great one—the disciplined difference between keeping up and leading the market. Core Expertise & Recent Innovations: LLM & RAG ArchitectureDesigned and implemented REF-RAG, a next-generation retrieval-augmented generation system featuring contextual compression and token-level semantic reweighting, significantly improving accuracy while reducing hallucinations.(Patent filing in progress) Enterprise AI OperationsBuilt an AI Operations Portal supporting SecOps, CloudOps, and NetOps—enabling intelligent ticket resolution, knowledge retrieval, RCA, and workflow orchestration using open-source LLMs and vector search. Autonomous AI & Agentic SystemsPioneered the second phase of our AI Ops platform, where autonomous agents execute validated remediation actions, moving toward self-healing network and security operations. High-Precision Semantic SearchIntegrated Weaviate Vector DB with hybrid/ Ensemble retrieval (dense + sparse) and multi-embedding models to deliver enterprise-grade semantic search across distributed knowledge bases. Broader ML & Data Science Impact: Algorithm InnovationDeveloped and implemented the K-Ball algorithm, improving neighbor detection accuracy at scale. Anomaly Detection & ForecastingBuilt ML-based anomaly detection systems that reduced manual monitoring by 50% and enabled proactive issue resolution. Technical Toolkit:Languages & Frameworks: Python, R, SQL, PySpark, TensorFlow, PyTorch, Scikit-learn, spaCy, Transformers, Ollama, Langchain, Vecor Database: Chroma / WeaviateMLOps & DevOps: Git, Flask, FastAPI, REST APIs, Docker, Docker-composeTools & Visualization: Linux, Nvidia GPU -Tesla T4 /A100, Tableau, Matplotlib, Seaborn, Plotly, pillow, Linux

Experience

1. Automation Genai Engineer

   Ericsson

   Jul 2025 — Present · Noida, IN

Education

• The University of Texas at Austin

  Postgraduate Program

  2021 — 2022

• North Maharashtra University

  Bachelor of Engineering - BE

  2005 — 2009