Sagar Ahire

Designed and deployed scalable agentic AI services using LangGraph, building stateful multi-agent orchestration pipelines with tool-calling, conditional routing, and human-in-the-loop workflows; integrated knowledge graphs to enable structured relational reasoning across agents for improved context retrieval in long-horizon tasks. Architected concurrent data processing systems using Python's ProcessPoolExecutor and async I/O, achieving measurable throughput improvements on 1M+ record datasets with low-latency guarantees. Developed secure RESTful APIs with FastAPI, Pydantic validation, Alembic migrations, and RBAC authorization — reducing production-related errors by 40%. Implemented a polyglot persistence layer (PostgreSQL + MongoDB) with optimized query strategies, supporting real-time analytics and high-volume ingestion workloads simultaneously. Optimized LLM token utilization across agentic pipelines through context pruning, prompt compression, and dynamic context window management — reducing token overhead and inference costs.

Designed and delivered a 3D Scanner Requisition System using ASP.NET, cutting internal tool request turnaround time by 35% across 4 cross-functional teams. Developed Python-based ML models to improve prediction accuracy and data consistency in manufacturing production workflows. Integrated Snap7 for PLC automation in collaboration with control engineering and robotics teams, directly enhancing manufacturing throughput.