Vidit Agarwal
AI ArchitectAI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI Architect with 10+ years in Agentic AI & MLOps Leadership
Principal AI Architect and GenAI Strategist with 13+ years of end-to-end experience designing, building, and scaling enterprise-grade AI systems across energy, finance, HR tech, ed-tech, and sports verticals. Specializes in Agentic AI, multi-agent orchestration, Retrieval-Augmented Generation (RAG), LLM pipeline engineering, and cloud-native MLOps. Proven track record of owning the full solution architecture lifecycle - from stakeholder requirement gathering and system design to production deployment, governance, and continuous improvement. Adept at translating ambiguous business problems into robust AI architectures that drive measurable ROI. Experienced leading distributed global teams and delivering complex AI programs for Fortune 500 and government clients including ADNOC, DEWA, and G20 organizations. Fully equipped for senior remote roles with reliable infrastructure and proven async delivery.
BITS Pilani
Master's in Artificial Intelligence & Machine Learning
N/A – Present
Ecosystem Consulting (Independent)
Principal AI Architect & GenAI Strategist
December 1, 2018 – Present
Global
Netway India BigFoot Retail Corwhite Solutions
Data Analyst / ML Engineer
March 1, 2015 – November 1, 2018
India
Idea Cellular Ltd
Data Analyst
March 1, 2013 – January 1, 2015
India
ADNOC GPT — Enterprise AI Platform
June 23, 2026 – Present
Owned end-to-end solution architecture for a ChatGPT-like enterprise AI platform for Abu Dhabi National Oil Company (ADNOC), enabling unified intelligence across finance, operations, procurement, and real-time news feeds. Designed a scalable, multi-modal data architecture integrating structured (ERP, SCADA, asset registries, finance) and unstructured (PDFs, reports, news) data into a unified vector + knowledge graph layer with semantic versioning and lineage tracking. Engineered multi-agent orchestration framework with specialized agents for research, summarization, cross-domain reasoning, and insight synthesis — combining finance, operations, and external market signals. Architected enterprise security, RBAC, and compliance layer on Azure OpenAI + AKS, with governance controls meeting ADNOC data sovereignty requirements. Led stakeholder alignment across C-suite, IT security, and domain SMEs; delivered phased rollout from PoC to production with change management and user adoption programs. Impact: Transformed fragmented enterprise data silos into a unified, real-time AI intelligence layer — enabling executive-level decision-making at a speed and depth previously impossible.
Tax.copilot - Enterprise Tax Intelligence Platform
June 23, 2026 – Present
Architected a production-grade, agentic AI platform for enterprise tax research using LangGraph multi-agent orchestration, Neo4j Knowledge Graph, and hybrid RAG pipelines combining dense vector search with graph traversal. Designed jurisdiction-aware multi-agent workflows encompassing query decomposition, intelligent retrieval, chain-of-thought reasoning, answer validation, and confidence scoring with explainable, auditable citations. Engineered a comprehensive enterprise knowledge base integrating Pagero regulatory data, Practical Law, internal client repositories, and real-time API enrichment — with automated ingestion, deduplication, and schema normalization. Implemented production-grade observability: LLM tracing, cost monitoring, latency SLAs, hallucination detection, and human-in-the-loop escalation pathways. Collaborated with legal and compliance teams to embed AI governance, audit trails, and explainability requirements into system design from day one. Impact: Reduced tax research time by 80%, transforming multi-hour manual workflows into a sub-minute AI-powered decision intelligence system — currently live in production.
Enterprise Document Intelligence Platform
June 23, 2026 – Present
Architected a fully agentic document processing platform handling Invoices, SOWs, Contracts, and Financial Reports using specialized LLM agents for OCR, multi-language translation, ERP field validation, cross-reference checks, and executive summarization. Engineered a RAG-Fusion pipeline with semantic retrieval, cross-encoder reranking, and context-aware financial Q&A systematically improving answer accuracy from 60% to 81% across benchmark test sets. Designed stateful agent memory and workflow checkpointing to enable long-running document workflows with pause-and-resume, exception handling, and SLA-aware routing. Built a self-correcting extraction loop leveraging LLM-based confidence scoring, structured output validation, and fallback to human review for edge cases. Impact: Processed 1,000+ documents/week; cut manual validation effort by 70% (10 hrs → 3 hrs per batch), slashed end-to-end processing time by 92% (2 days → 2 hours), and achieved 98% factual accuracy on summarization benchmarks.
Autonomous Invoice Reconciliation Platform
June 23, 2026 – Present
Designed and delivered a production-grade reconciliation system for global cross-border shipment invoices using LangGraph stateful agents + RAG, replacing brittle hand-coded regex pipelines with resilient LLM-driven structured extraction. Architected explainable mismatch detection with configurable severity scoring, tolerance rules, and multi-format (EDI, PDF, XML, CSV) cross-system validation across ERP and logistics platforms. Implemented intelligent exception triage: mismatches auto-classified by root cause, routed to the correct downstream team, and tracked through resolution with full audit trail. Impact: Eliminated weeks of manual reconciliation backlog; reduced dispute resolution cycle time by 60% and false-positive exceptions by 45%.
ObeAssess (EdTech • GCP)
June 23, 2026 – Present
AI assessment platform on GCP Vertex AI with multi-agent RAG, LLM-based essay scoring, and multilingual support (Arabic/English) — scalable to 100K+ concurrent students.
Sports Analytics Agent
June 23, 2026 – Present
Multi-agent NL-to-SQL system using LangGraph, ChromaDB, Neo4j, GPT-4/DeepSeek — schema-aware with subquery planning, semantic reranking, and self-correction loops.
GenAI Power Generation Platform
June 23, 2026 – Present
Vertical GenAI for SCADA/IoT operational data — RAG + LLM for anomaly interpretation with Explainable AI (XAI); delivered 5× faster root-cause analysis for plant engineers.
DEWA Proposal Evaluation
June 23, 2026 – Present
GPT + RAG + Neo4j knowledge graph for automated vendor scoring and proposal evaluation — reduced evaluation cycles by 45% while improving consistency and auditability.
Email Workflow Agent
June 23, 2026 – Present
LLM-powered email intelligence agent for intent parsing, sub-task decomposition, and downstream workflow automation — reduced query resolution effort by ~40%.
G20 AI Reporting Suite
June 23, 2026 – Present
Automated multilingual reporting pipeline for G20 policy documents using LangChain + RAG, enabling rapid cross-domain synthesis across 10+ policy domains.
Databricks Certified ML Professional
Databricks
June 1, 2026 – Present
Google Cloud Professional Data Engineer
Google Cloud
June 1, 2026 – Present
Microsoft Azure AI Fundamentals
Microsoft
June 1, 2026 – Present
Cultural Fit Analysis
The candidate's diverse project portfolio across energy, finance, HR tech, ed-tech, and sports verticals, combined with experience serving government and Fortune 500 clients globally, indicates a high degree of adaptability and cultural awareness. Their role as an independent consultant further emphasizes self-reliance, initiative, and a broad understanding of different organizational needs. The breadth of technical skills and exposure to various cloud platforms and AI frameworks suggests a continuous learning mindset, which is vital for cultural fit in an innovative AI environment.
Soft Skills & Operational Fit
The candidate demonstrates strong leadership, stakeholder management, and cross-functional delivery skills, crucial for an AI Architect role. Their experience with remote work and async communication, coupled with a proven track record of delivering complex AI programs globally, indicates excellent operational fit for distributed teams. The detailed project descriptions highlight a problem-solving mindset and an ability to drive significant business impact.