Sanjay Yadav

Generative AI Engineer / GenAI Developer with 2.6 years at TCS building enterprise RAG systems, LLM-based classification pipelines, agentic AI applications, and enterprise Python automation. Hands-on with LangChain, LangGraph, MCP, vector databases (Milvus, FAISS), OpenAI APIs, and SQL — including a production RAG + LLM incident-classification pipeline (90% classification accuracy) and a multi-agent travel-planning system with human-in-the-loop workflows. AWS, Databricks, and SAP certified in Generative AI.

GenAI Developer (System Engineer) at Tata Consultancy Services (TCS) (2024-01 – Present)

Chronic Incident Resolution Initiative (CIRI) — Enterprise RAG for Incident Automation. Tech Stack: Python, LangChain, OpenAI (GPT-4o-mini), OpenAI Embeddings, Milvus (PyMilvus), RAG, ServiceNow, Snowflake, SQL, ignio

• Designed and implemented CIRI, an AI-driven RAG framework that automatically classifies and resolves repetitive access-related ServiceNow incidents using OpenAI embeddings, GPT-4o-mini, and Milvus-based semantic similarity search over historical incident records.
• Architected the vector database schema and embedding strategy in Milvus for efficient retrieval of incident knowledge, and integrated ServiceNow, Snowflake, and ignio into a single end-to-end automated resolution workflow.
• Reduced LLM token consumption and operational cost by adopting a RAG-first approach (retrieve-then-classify instead of sending every incident directly to the LLM) and migrating from GPT-4o to GPT-4o-mini.
• Built a supervised-learning feedback loop that continuously improves classification accuracy using resolved-ticket history and worked with support teams and domain experts on resolution patterns, prompt design, and production-readiness governance, monitoring, and QA.
• Achieved ~90% classification accuracy in production, saving approximately 400 manual hours per year for one account.

GenAI Developer (System Engineer) at Tata Consultancy Services (TCS) (2024-01 – Present)

IDoc Alert Monitoring — Python / Robocorp Automation. Tech Stack: Python, Robocorp, SAP

• Built a Python/Robocorp automation that continuously monitors SAP IDocs for failures across multiple message and status types, with condition-based alerting that notifies relevant users when a failed IDoc matches defined status/severity criteria.
• Reduced approximately 360 manual hours per year by replacing manual IDoc monitoring with automated identification and notification.

GenAI Developer (System Engineer) at Tata Consultancy Services (TCS) (2024-01 – Present)

Delivery Tie-Out Report — Ignio Automation. Tech Stack: ignio, SAP, Data Processing, Reporting

• Built an ignio-based automated workflow to extract, reconcile, and deliver the Delivery Tie-Out report from multiple SAP tables without manual intervention.
• Reduced approximately 420 manual hours per year and improving data reliability.

GenAI Developer (System Engineer) at Tata Consultancy Services (TCS) (2024-01 – Present)

Montend Flat File Upload — Python / Robocorp Automation. Tech Stack: Python, Robocorp, SAP, File Processing

• Built an automation that validates and uploads user-submitted flat files via an SAP process chain, with automated status monitoring and confirmation alerts.
• Reduced approximately 24 manual hours per year.

B.Tech in Computer Science and Engineering – Rajkiya Engineering College, Sonbhadra (2019-07 – 2023-06)