Kailas Shingate

AI Software Engineer specializing in the design and deployment of secure, production-grade LLM orchestration pipelines and multi-tenant SaaS platforms. Expert in building stateful, multi-agent systems (LangGraph), zero-trust Data/RAG architectures, and asynchronous API backends. Passionate about bridging the gap between complex enterprise databases and natural language by engineering fault-tolerant AI agents that drive automated business intelligence and dynamic data visualization

AI Engineer at Self Directed Project (2025-10 – Present)

Live URL - https://datapilotai-ndt4k9thd8rmet6zxlh464.streamlit.app/

• Multi-Tenant SaaS Architecture: Engineered a secure, enterprise NLP-to-SQL platform using an asynchronous FastAPI backend and PostgreSQL pool management (Asyncpg), supporting isolated database environments for multiple organizational tenants.
• Agentic Orchestration (LangGraph): Designed a stateful multi-agent pipeline with a dual-model routing architecture (Gemini Pro/Flash & Groq Llama 3). Implemented an autonomous retry node that feeds database execution errors back into the LLM for self-correction, vastly improving query success rates.
• Advanced RAG & Memory Management: Built a self-healing schema indexing engine using Pinecone. Developed a custom 'Garbage Collection' algorithm using Set Theory to hunt and purge 'ghost vectors' from AI memory when client database tables are deleted. Implemented SHA-256 metadata hashing to detect data drift and skip redundant embedding generation, saving LLM token costs.
• Enterprise Security & RBAC: Architected a zero-trust execution environment featuring JWT-based Role-Based Access Control (RBAC). Secured tenant database connections using AES-256 Fernet encryption, and enforced read-only PostgreSQL execution constraints to guarantee client data safety.
• Dynamic UI & Visualization Router: Developed a stateless, Dribbble-styled Streamlit frontend. Built an autonomous charting engine that forces the LLM to output strict Pydantic/JSON schemas, dynamically mapping complex SQL results to interactive Plotly visualizations without executing raw Python code.

Project Engineer at Wipro (2022-05 – 2024-11)

• Designed, developed, and maintained OBIEE Repository (RPD) layers (Physical, Business Model, and Presentation) to support enterprise-wide reporting and analytics.
• Developed interactive dashboards in OBIEE to visualize key performance metrics, enabling data-driven decision-making for clients.
• Provided technical support for OBIEE Repository-related issues, resolving 95% of tickets within SLA timelines.
• Documented and maintained RPD design specifications and troubleshooting guides, reducing resolution time for recurring issues by 20%.

PG Program in Data Science, Machine Learning & Neural Network – Data Trained Education Pvt. Ltd. (2023-01 – 2024-01)

BE in Electronics and Telecommunication – Pune University (2018-08 – 2021-07)