AI Solution Architect - Anthropic
(2025-11)
Agentic AI architecture for regulated industries — Anthropic Claude, Amazon Bedrock. Flagship engagement: Agentic AI Transformation for Auto Loan Origination & Servicing (regulated financial services). Leading design of an end-to-end agentic AI solution for a leading auto loan processor, with Anthropic Claude as the core reasoning engine, automating and orchestrating the complete loan origination and servicing lifecycle.
- Partnered with business stakeholders, loan officers, underwriters, and compliance teams to map the full auto loan value chain — documenting current-state workflows, decision rules, exception paths, and regulatory checkpoints.
- Identified and prioritized high-value agentic automation use cases by processing volume, manual effort, error rates, and customer experience impact; decomposed end-to-end processes into well-bounded tasks for autonomous agent execution with defined inputs, outputs, success criteria, and escalation triggers.
- Designed governance and observability from day one: prompt evaluation frameworks, human-in-the-loop review consoles, audit logging, and responsible AI guardrails suited to regulated lending.
- Stack: Anthropic Claude, Claude Agent SDK, Model Context Protocol (MCP), tool use / function calling, multi-agent orchestration patterns, REST APIs, event-driven workflows, document AI pipelines, core banking and LOS integrations.
Director — AI and Analytics Solutions - HCL Technologies (America)
(2024-12 - 2025-10)
Enterprise AI Transformation for a Regulated Power Distribution Utility (AI Architect). Partnered with CXO stakeholders to define the enterprise AI strategy and architecture roadmap for a regulated utility.
- Architected and curated a portfolio of 100 AI use cases across grid operations, asset management, field operations, customer experience, and regulatory compliance, prioritized via a value-vs-feasibility framework.
- Designed the reference architecture for on-premises GenAI deployment using Meta's Llama LLM, addressing data sovereignty, cybersecurity, and NERC/PUC regulatory requirements.
- Led end-to-end architecture and delivery of two strategic PoCs — Field Technician Copilot (RAG-based assistant integrating SCADA, outage history, and SOPs to accelerate fault detection, diagnostics, and outage response) and Regulatory Correspondence Assistant (AI solution to ingest, classify, and draft responses to government and Public Utility Board requests, mapped to policies and prior submissions).
AI/ML Solution Architect - AWS
(2022-04 - 2024-11)
- Project: AI-Powered Email Classification (AWS Comprehend, Bedrock, Lambda, S3, RDS). Designed and implemented a fully automated architecture ingesting, analyzing, and classifying incoming email into categories such as Support, Sales, Billing, and Spam. Delivered an 80% reduction in manual email triage and raised classification accuracy to 95% by combining AWS Comprehend NLP (entities, sentiment, topics) with Amazon Bedrock foundation-model LLMs for contextual understanding. Built event-driven, s
- Project: AI-Powered Stock Price & Volume Anomaly Detection and Visualization. Architected and deployed a real-time anomaly detection and visualization system identifying unusual price fluctuations and trading-volume spikes in financial markets for institutional traders. Achieved 90%+ anomaly detection accuracy with sub-second latency, and cut infrastructure costs by 60% via a serverless-first architecture. Trained time-series models (Isolation Forest, AutoML) on Amazon SageMaker; integrated Lamb
- Stack: AWS Lambda, SageMaker, RDS, S3, DynamoDB, QuickSight, EventBridge, SNS, Step Functions, Terraform, CloudWatch, Python (Boto3, Pandas, Scikit-Learn, XGBoost, Statsmodels), SQL.
GCP Solution Architect — Analytics, AI and ML - Tata Consultancy Services
(2020-11 - 2022-03)
Project: AI-Driven Enterprise Application Migration to Google Cloud for AI/ML Enablement. Led migration of 16 enterprise applications from on-premises and legacy cloud environments to Google Cloud Platform, unlocking AI/ML capability across business functions.
- Improved application performance and scalability by 40%, cut infrastructure and operational costs by 60% via GCP serverless and managed services, and increased deployment speed 3x with automated CI/CD pipelines and Infrastructure as Code.
- Enabled enterprise AI/ML adoption — predictive analytics, automated workflows, and intelligent decision-making — on Vertex AI, AutoML, TensorFlow, BigQuery ML, and Dataflow.
- Stack: Vertex AI, AutoML, TensorFlow, GKE, Cloud Run, Cloud Functions, BigQuery, Firestore, Cloud Spanner, Dataflow, Terraform, Cloud Build, GitOps.
Data Analytics Consultant
(2018-03 - 2020-10)
Project: AI-Powered Predictive Analytics for Package Delivery Optimization (leading logistics company).
- Improved on-time deliveries by 25% through predictive delay forecasting and reduced operational inefficiencies by 30% via AI/ML-optimized allocation of delivery personnel and equipment.
- Enabled real-time streaming analytics for instant operational decisions, improving customer satisfaction and logistics planning; drove fuel and idle-time savings via AI-driven route and resource optimization.
- Stack: Amazon SageMaker, AWS Glue, Redshift ML, Kinesis, Lambda, Apache Spark, S3, EC2, Step Functions, CloudFormation, Pandas, Scikit-learn.
Senior Managing Consultant — Analytics Center of Competency - IBM
(2015-09 - 2018-02)
Project: AI-Powered Predictive Analytics for Drilling Blowout Prevention (oil & gas). Reduced blowout incidents by 40% with real-time AI-driven predictive alerts, improving drilling safety, minimizing non-productive time, and cutting operational costs through predictive maintenance. Built ML models in IBM SPSS Modeler on historical drilling data, well parameters.
