Manoj Kumar Karumanchi

Software AI Engineer specialized in applied AI, RAG systems, agentic workflows, and LLMOps.

Experience building production-grade AI pipelines across Azure, GCP, and AWS. Strong focus on workflow automation, latency reduction, and scalable multimodal systems. Skilled in designing end-to-end AI components and improving platform reliability.

Software Engineer at Accellor (2024-07 – Present)

Built production-grade AI pipelines and agentic systems across multiple cloud platforms

• Built an automated media-processing pipeline using S3 triggers, Lambda, ECS, Textract, Transcribe, and Comprehend to extract text, audio, and metadata into structured outputs such as summaries and process flows
• Built and deployed a multimodal Agentic RAG platform using Azure AI Document Intelligence + Azure OpenAI, enabling automated research reports and PPT generation with advanced chunking and workflow orchestration
• Developed a document processing pipeline (AWS Lambda, ECS, S3) automating large-scale document ingestion with 90% accuracy, integrating Azure Document Intelligence and OpenAI for multimodal extraction
• Developed a Conversational Analytics Platform (LangChain + Azure OpenAI) supporting intent classification, automated SQL generation, and natural-language data exploration; integrated D3.js full-stack visualizations
• Reduced voice agent workflow latency by 40% by deploying real-time MCP-based recognition and automation
• Developed and deployed an internal HR assistant using GCP Vertex AI (Gemini), automating recurring HR queries and reducing HR workload by 60% while cutting employee query resolution time by 70%
• Published PyPI packages integrating MCP server capabilities for Claude and Cursor, enabling faster AI agent development
• Prototyped a unified pipeline using Agent Bricks with Delta Lake ingestion, feature processing, MLflow-driven prediction workflows, and automated Genie AIBI reporting with full governance and observability

Data Analyst Intern at Senchola Technologies (2023-06 – 2023-08)

Data analysis and machine learning project work

• Improved flight delay prediction accuracy by 30% using Random Forest + XGBoost on web-scraped datasets
• Reduced spam false positives by 30% with a custom NLP classifier and deployed a Dialogflow-based chatbot
• Delivered 95%+ accuracy across ML models for wine quality prediction, car price forecasting, and LSTM-based stock models

Bachelor of Technology in Computer Science and Engineering – Indian Institute of Information Technology, Tiruchirappalli (2020-11 – 2024-06)