Sid Karlekar

About Sid Karlekar

I build production AI systems and modern data platforms that operate under real-world constraints. latency, scale, cost, and compliance and not just demos.With 15+ years across fintech, insurance, and enterprise systems, I’ve worked end-to-end:From legacy ETL (SSIS, SQL Server, batch pipelines)To cloud-native data platforms (Kafka, Snowflake, Databricks)To AI-native systems (RAG, agentic workflows, LLM platforms)What I Do:Build production RAG systems with vector search, reranking, and observabilityDesign agentic AI workflows (LangGraph, tool-calling, multi-step reasoning)Architect real-time streaming platforms (Kafka + Snowflake/Redshift)Develop AI-powered data validation and anomaly detection systemsBuild LLM copilots replacing dashboards and manual analysis-Data Engineering & Platform Expertise:Event-driven architectures (CDC, Kafka, streaming pipelines)Modern ELT platforms (Snowflake, Databricks, dbt)Data quality, observability, and cost optimizationSchema evolution and high-scale ingestion pipelinesMigrating legacy systems into cloud-native, AI-augmented architectures-AI & LLM Focus:RAG (Retrieval-Augmented Generation)Agentic systems (LangGraph, orchestration, tool use)LLM platforms (Claude, GPT-4, Llama, Mistral, Bedrock, Azure OpenAI)Embeddings, vector search, reranking, and retrieval optimizationLLM evaluation, monitoring, and guardrailsTech Stack-AI / LLM / AgentsLangChain • LangGraph • LlamaIndex • Agentic Workflows • Tool Calling • Prompt Engineering • Fine-Tuning (LoRA/QLoRA) • Claude • GPT-4o • Llama • Mistral • Hugging Face • MLflow-Data EngineeringKafka • Spark • PySpark • Snowflake • Databricks • Redshift • PostgreSQL • SQL Server • CDC • ETL • ELT • dbt • Airflow • Delta Lake-Cloud & InfrastructureAWS (Lambda, S3, SSM, Bedrock, Glue) • Azure (Azure OpenAI, Data Factory, Synapse, Azure Functions, DevOps) • Terraform • Docker • CI/CD• Schemachange • Debezium-Observability & AI OpsMLflow • CloudWatch • Dynatrace • LLM Monitoring • Prompt Monitoring • Data Quality Frameworks• Datadog-Leadership & ImpactLed teams delivering AI + data systems across organizationsMentored engineers on AI engineering, data pipelines, and platform designPresented AI architecture and strategy to leadership and stakeholdersBuilt systems impacting revenue, operations, and decision-making-Current FocusScaling production-grade RAG + agentic systemsBuilding AI-native data platforms replacing traditional ETL pipelinesExploring autonomous systems combining data + reasoning

Experience

1. Senior Cloud Data Engineer

   Orion

   Jan 2025 — Present · Omaha, NE, US

   AI Systems-Architected and deployed ServiceBot, a production RAG assistant serving internal operations teams and financial advisors across Microsoft Teams and Glia under real-world traffic and compliance constraints-Implemented contextual retrieval (Anthropic technique), significantly improving retrieval accuracy and answer relevance over baseline embedding approaches-Built end-to-end knowledge ingestion pipeline (3,500+ documents) including OCR extraction, PII masking, semantic chunking, metadata tagging, and vector indexing (Databricks Vector Search, HNSW)-Designed two-stage retrieval architecture (candidate retrieval → domain filtering → cross-encoder reranking → LLM response generation)-Integrated Claude (Databricks Model Serving) with custom embedding pipelines, managed via MLflow and Unity Catalog-Developed agentic AI workflows using AWS Bedrock, enabling tool-calling against internal financial APIs for real-time client intelligence-Designed multi-agent orchestration patterns for tool-augmented LLM systems and autonomous workflowsData Platform & Engineering-Designed and scaled real-time streaming pipelines (SQL Server → Kafka → Snowflake/Redshift) using AVRO schemas for high-volume, low-latency ingestion-Built Change Event Log (CDC) architecture using triggers, stored procedures, and event-driven ingestion patterns-Led migration from legacy ETL (SSIS, batch pipelines) to modern ELT + AI-augmented architectures (Databricks, Snowflake, dbt-ready patterns)-Integrated LLM-powered anomaly detection into streaming pipelines using LangChain + OpenAI-Designed Snowflake-native cost attribution engine for multi-tenant compute and storage optimizationLeadership & Impact-Led architecture discussions aligning AI + data platform strategy with business and product goals-Acted as a bridge between data engineering, AI systems, and business stakeholders

Education

• Medi-Caps Institute of Technology & Management

  Bachelor of Engineering - BE, Computer Science

  2007 — 2010

• University of Nebraska at Omaha

  Master’s Degree, Management Information Systems, Data Analytics

  2016 — 2017