Ashutosh Rawat

Architected a distributed GenAI conversational platform using a 4-microservice architecture with dedicated services for REST APIs, WebSocket communication, LLM agent orchestration with vector retrieval, and automated knowledge-base generation. Designed core extraction algorithms and implemented end-to-end pipelines to process structured and unstructured Excel, PDF, and image data for RAG-based retrieval systems. Built an intelligent knowledge-base creation engine integrating OCR, structured parsing, embeddings, chunking, and metadata enrichment to support high-accuracy RAG workflows.

R&D Engineer I at VassarDigital.ai (2024-12 – Present)

• Architected a distributed GenAI conversational platform using a 4-microservice architecture with dedicated services for REST APIs, WebSocket communication, LLM agent orchestration with vector retrieval, and automated knowledge-base generation.
• Designed core extraction algorithms and implemented end-to-end pipelines to process structured and unstructured Excel, PDF, and image data for RAG-based retrieval systems.
• Built an intelligent knowledge-base creation engine integrating OCR, structured parsing, embeddings, chunking, and metadata enrichment to support high-accuracy RAG workflows.
• Engineered asynchronous, parallel, event-driven processing pipelines using Kafka/Azure Event Hub, reducing system latency by 50% and improving throughput across large-scale document batches.
• Implemented hybrid semantic retrieval using Milvus + ChromaDB + Knowledge graph retrieval, improving contextual relevance and retrieval Accuracy by 30%.
• Led development of Production Planning Optimization for Kavach equipment (Indian Railways project) manufacturing, generating optimized daily and weekly schedules to meet demand deadlines while minimizing production costs, reducing manual operational effort by 80%.
• Integrated Microsoft Graph API to deploy a Teams chatbot, enabling seamless conversational access to backend services directly within Teams.

R&D Intern at VassarDigital.ai (2024-06 – 2024-11)

• Built an AI voice assistant using Twilio + Deepgram supporting real-time transcription, intent switching, and context-aware dialogues for hospital appointment booking.
• Created APIs for an admin dashboard to monitor chat analytics, including metrics on resolved and unresolved conversations. Wrote unit tests improving API reliability and coverage.

Bachelor of Technology in Computer Science Engineering – Indian Institute of Information Technology, Sri City AP, India (2020-12 – 2024-05)