Bharat Prajapat

About Bharat Prajapat

I\'ve saved enterprises $28M. Slashed AI latency by 92%. Shipped multi-agent systems to production while others are still in pilot. With 8+ years progressing from Cloud Engineer → DevOps → SRE → Platform Engineering → Agentic AI Architect at IBM, I\'ve done what most AI engineers only talk about: I\'ve moved Fortune 500 companies from \'AI demo\' to \'AI profit\' — with real numbers on the board. WHAT I BUILD Multi-agent orchestration systems — autonomous workflows that think, plan, and execute at scale LLMOps & MLOps pipelines — Kubeflow, MLflow, Vertex AI, SageMaker, GitOps Distributed ML inference — GPT-4, LLaMA 3, BERT; 92% latency reduction via quantization & batching RAG pipeline infrastructure — Pinecone, FAISS, vector databases at enterprise scale AI Governance & Compliance — SOC 2, GDPR, EU AI Act, Zero Trust, bias mitigation Cloud-native AI platforms — GKE, EKS, AKS, Kubernetes, Terraform, hybrid multi-cloud. IMPACT THAT SPEAKS → $28M+ projected savings from agentic workload optimizer + 280K ton CO₂ reduction → 92% AI inference latency reduction across multi-cloud LLM deployments → 90% faster model deployment via enterprise LLMOps pipelines → $250K+ annual FinOps savings: serverless AI + GPU auto-scaling → 75–90% deployment time reduction across 4 Fortune 500 clients CLIENTS I\'VE DELIVERED FOR Mercedes-Benz (Germany) · TD Bank (Canada) · Capital One (USA) · American Express (USA) · Lygon (AU) TECH ARSENAL Agentic AI: IBM watsonx, LangChain, ReAct agents, tool-calling, guardrails, autonomous workflows LLMOps: Kubeflow Pipelines, MLflow, Vertex AI, SageMaker, Azure ML, Triton Inference Serve Cloud: GCP (Vertex AI, GKE, BigQuery ML, Dataflow, Pub/Sub) · AWS · Azure Infrastructure: Kubernetes, Terraform, Docker, Helm, GitHub Actions, Prometheus, Grafana Data: Pinecone, FAISS, Kafka, ElasticSearch, PostgreSQL, Redis, Snowflake. b••••••••@gmail.com GitHub: github.com/bharatxnj Credly: credly.com/users/bharat-prajapat

Experience

1. Ai Platform & Agentic Ai Architect Llmops, Governance & Multi-cloud Leader

   IBM

   Aug 2024 — Present · Bengaluru, IN

   Architected production-grade multi-agent orchestration systems on IBM watsonx Orchestrate with ReAct reasoning loops, autonomous tool-calling, and LLM guardrails — serving millions of enterprise workflow events daily for Fortune 500 clientsDeployed GPT-4, LLaMA 3, and BERT at scale on Google Vertex AI, AWS SageMaker, and Azure ML; achieved 92% inference latency reduction via model quantization, dynamic batching, and Triton Inference Server-style GPU routingEngineered end-to-end LLMOps pipelines using Kubeflow Pipelines, MLflow, Terraform, and GitOps workflows — compressed model deployment cycles by 90% from weeks to hours; enabled continuous model evaluation, A/B testing, and automated rollbackBuilt agentic workload optimizer using distributed ML scheduling across GPU/CPU hybrid clusters — projected $28M+ enterprise savings and 280K ton CO₂ reduction through intelligent multi-cloud resource allocation and spot instance arbitrageDelivered $250K+ annual savings via AI FinOps: serverless inference endpoints, GKE-style GPU auto-scaling, right-sizing workloads, and spot/on-demand arbitrage across AWS, Azure, and GCPLed AI governance & compliance program: SOC 2, GDPR, EU AI Act readiness, Zero Trust security, model explainability, bias detection pipelines, and full audit trails for regulated financial services and automotive clients in UK, Germany, and CanadaBuilt RAG pipeline infrastructure using Pinecone and FAISS vector databases; implemented LLM observability with Prometheus/Grafana for real-time model monitoring, data drift detection, and automated retraining triggers in productionIntegrated BigQuery ML for feature engineering, Pub/Sub for event-driven inference, and Dataflow for real-time ML data processing — building the data backbone for production AI at enterprise scale

Education

• Devi Ahilya Vishwavidyalaya

  Bachelor's degree, Information Technology

• Chicago State University

  Master's degree, Computer Science