01
Agentic AI systems
Hierarchical orchestration, autonomous planning, tool routing, memory, and human-in-the-loop workflows.
Agentic AI Engineer for enterprise-scale autonomous systems
Production-grade agentic AI for enterprise execution.
Over 4 years of experience designing and deploying enterprise AI systems that combine LLM orchestration, retrieval-augmented generation, agent memory, workflow automation, and human-in-the-loop controls.
Mrunalsingh Zire
Pune, Maharashtra, India / Accenture Agentic AI Specialist
Focus
Multi-agent orchestration, enterprise RAG, agent memory, workflow automation, and governed GenAI platforms.
Currently Building
- Persistent agent memory systems
- Eval-first release workflows
- Governed enterprise automation at scale
01 / Dossier
Enterprise agentic AI engineer with 4+ years in production systems.
Specialized in autonomous systems, hierarchical multi-agent architectures, enterprise retrieval, evaluation, governance, and large-scale AI integration across AWS, Azure, and Kubernetes environments.
I build reliable autonomous systems that solve enterprise work at scale.
My work combines orchestration, contextual reasoning, tool execution, retrieval grounding, and enterprise integration so AI systems can plan, act, recover, and remain observable in production.
I focus on measurable business outcomes: reducing operational costs by 30%, improving MTTR by 15%, scaling platforms for 500K+ users, and improving workflow continuity through stronger memory and evaluation.
02
LLM and RAG engineering
GPT-4, Claude, Llama, hybrid retrieval, embeddings, grounding, hallucination mitigation, and semantic search.
03
Infrastructure and backend
Python, FastAPI, distributed services, ServiceNow integrations, event-driven systems, AWS, Azure, and Kubernetes.
04
Evaluation and governance
DeepEval, TruLens, observability, drift detection, safety guardrails, explainability, and enterprise governance.
02 / Work
Enterprise AI delivery highlights with measurable impact.
Production systems spanning ITSM automation, enterprise knowledge retrieval, multilingual AI, and scalable AI platform delivery.
Case 01
500K+ enterprise users supported
Amethyst Multi-Agent Automation Platform
Enterprise multi-agent platform for ITSM operations, incident handling, SLA monitoring, escalation, root cause support, and governed enterprise knowledge retrieval.
LangGraph
CrewAI
GPT-4
FastAPI
ServiceNow
Kubernetes
Architected 3 orchestration agents coordinating 11 specialized utility agents with asynchronous task delegation, persistent memory, contextual reasoning, and bi-directional ServiceNow integrations.
85%autonomous completion
92%operational accuracy
30%cost reduction
Case 02
100K+ queries per month
Enterprise Context-Aware AI Assistant
Context-aware conversational AI for enterprise knowledge access using hybrid retrieval, reranking, query expansion, and low-latency answer generation.
Pinecone
BM25
RAG
Reranking
AWS Lambda
FastAPI
Built low-latency inference pipelines with Pinecone plus BM25 retrieval, secure API orchestration, source grounding, and scalable microservice deployment while keeping p95 latency below 2 seconds.
35%relevance lift
50%fallback reduction
<2sp95 latency
Case 03
10M+ documents searchable
GenAI Knowledge Management System
Enterprise knowledge assistant using AWS Bedrock for natural-language search, multilingual retrieval, trust scoring, and contextual answers across large document repositories.
AWS Bedrock
Embeddings
Multilingual RAG
Confidence Scoring
Citations
Implemented source attribution, confidence scoring, citation tracking, interactive follow-ups, and support for 12+ languages across 10M+ enterprise documents.
4.7/5satisfaction score
12+languages supported
10M+documents searchable
03 / AI Lab
Core expertise across the agent, retrieval, and infrastructure stack.
From orchestration and retrieval to deployment, observability, governance, and enterprise AI delivery.
Capability areas
Agentic AI and orchestration
Multi-agent architectures, hierarchical planning, tool calling, context management, and inter-agent communication.
Core
LLM and retrieval engineering
RAG, hybrid dense-sparse retrieval, reranking, prompt engineering, semantic search, and grounding.
Core
Cloud and backend systems
Python, FastAPI, distributed microservices, ServiceNow integrations, API design, and scalable AI backends.
Core
Evaluation and governance
DeepEval, TruLens, observability, drift detection, guardrails, explainable AI, and governed enterprise deployment.
CoreStack map
Tools and platforms used to build reliable, scalable, and production-ready enterprise AI systems.
GPT-4
Claude
Llama
LangGraph
CrewAI
OpenAI APIs
Pinecone
FAISS
BM25
Hybrid Retrieval
Re-ranking
Python
FastAPI
Flask
ServiceNow
AWS Bedrock
SageMaker
Lambda
Azure
Docker
Kubernetes
DeepEval
TruLens
CI/CD
04 / Credentials
Education, certifications, and professional focus.
A concise view of the academic foundation, recent certifications, and technical domains shaping the next phase of my work.
Education2017 - 2021
B.E. in Computer Science and Engineering
G.H. Raisoni College of Engineering, with a foundation in scalable software systems, applied AI, and engineering fundamentals.
Computer Science
Engineering
Certifications2024
Generative AI certifications
Completed Generative AI with Large Language Models from DeepLearning.AI and Generative AI for Everyone from Google.
DeepLearning.AI
Google
InterestsOngoing
Autonomous systems, reliability, and governed AI
Focused on agentic workflow architectures, enterprise GenAI platforms, AI reliability, human-AI collaboration, and responsible AI systems.
Autonomous AI
Responsible AI
05 / Contact
Open to high-impact enterprise AI engineering work.
Interested in roles focused on agentic AI platforms, enterprise GenAI architecture, autonomous workflows, AI reliability, and production-scale LLM systems.
Email
mrunalsinghzire@gmail.com
Phone
+91-8208396310
Location
Pune, Maharashtra, India
Focus
Agentic AI / Multi-Agent Orchestration / Enterprise RAG
Current Role
Full Stack LLM Development Analyst
Approach
Reliable, observable, governed, production-ready AI systems