Dipjyoti Das

AI Architect with 13+ years delivering production AI/ML/GenAI systems across startups and Fortune 150 companies — spanning Technology, Finance, Healthcare, Utilities, Insurance, Telecom, and Logistics

Generative AI & Agentic AI — Enterprise Impact

Charter Communications — Led architecture and deployment of a unified Agentic AI chatbot microservice on AWS EKS, serving 80,000+ field technicians and engineers. Reduced support calls by 60% with million-dollar savings.

CareSource — Led an Agentic AI solution with LangGraph to automate clinical discharge log fax intake process using Azure Document Intelligence and LLM validation with human-in-the-loop review. Integrated eligibility and Prior Authorization APIs to reduce manual review and improve auditability.

Fannie Mae — Served as AI Architect during Fannie Mae’s foundational Enterprise GenAI journey on AWS Bedrock/SageMaker. Defined roadmap and milestones across business units; built knowledge assistants and reusable accelerators. Led POC for Agentic AI automating Foreclosed Property Valuation — reducing approval time by 70% through multi-agent orchestration for document ingestion, validation and red-flag detection.

CGI — Pulse AI Platform — Built core Agentic AI capabilities within CGI Pulse AI, an enterprise GenAI product serving CGI’s global clients. Designed LangGraph ReAct agents using Reflection, Planning and Multi-Agent patterns, and built the no-code Agent configuration console with full observability with Langfuse — deployed as self-hosted microservices on Azure Kubernetes Service.

AT&T — Developed multi-agent workflows within AT&T’s Ask ATT platform for customer service automation, architecting a Team Planner Agent orchestrating sub-agents with a Validator Agent for output quality and compliance.

Traditional AI/ML — Proven at Scale

• $240M incremental annual revenue — Propensity model for annuity product targeting of Financial Advisors at Brighthouse Financial, deployed to SAP cloud with PySpark pipelines.
• $10M/year in savings — Risk Propensity and Energy Efficiency recommendation models at Duke Energy for business market team.
• $800K/year saved — Automated time-series forecasting at R+L Carriers for sales operations.
• 30M properties scored — Resilience and Business Interruption analytics at One Concern, integrated into Swiss RE CatNet and Willis Tower Watson globally.

I work across the full AI delivery lifecycle — hands-on architecture and coding, team leadership, and executive advisory — across Marketing, Sales, Product, Operations, and Engineering functions at startups and large enterprises.

I am passionate about moving AI from prototype to production — designing systems that are robust, observable, and actually useful in the real world.

Areas of expertise: Agentic AI · RAG systems · LLMOps · Generative AI Architecture · MLOps · · Python · LangGraph · LangChain · Cloud (AWS, Azure, GCP) · Kubernetes · Serverless

---

Let’s Connect

If you have any questions about Gen AI, Agentic AI or Machine Learning, or want to exchange ideas, feel free to reach out.