Experience

Gen AI Architect — Ascendion, Charlotte NC

Oct 2025 – Present

Ascendion is a software engineering firm delivering enterprise Generative and Agentic AI solutions. Led architecture and production deployment of two large-scale AI systems across Telecom and Healthcare enterprises.

• Led cross-functional engineering teams to Architect and Deploy a unified Agentic AI chatbot microservice on AWS EKS within Charter Communications’ Tech Mobile application, serving 80,000+ field technicians, supervisors, maintenance engineers.
• Architected LangGraph-based orchestration to route role and intent aware queries across task-specific agents (Job Insight, Tech Assist, APIs) with Langfuse observability; reduced support calls by 60%, accelerated service completion with million-dollar savings.
• Architected stateless REST services with FastAPI, OAuth/JWT security, and PostgreSQL/Redis-backed state management to support scalable enterprise workloads.
• Led development of a Construction Agent RAG service by extracting and embedding 1,000+ confluence pages and 5,000+ technical attachments (images, pdfs, excel, word) into PGVector DB, preserving document hierarchy and multi-level contextual relationships.
• Design and lead an Agentic AI solution (LangGraph) for CareSource Healthcare client to automate Discharge logs fax intake process using Azure Document Intelligence and LLM validation (HITL), integrating eligibility and Prior Authorization APIs to reduce manual review; improve auditability.
• Led serverless deployment with Azure Functions & CI/CD; implemented observability via Application Insights and centralized prompt versioning, logging. Oversaw React-based dashboard deployed on Azure Web App to enable business review & HITL validation.
• Owned end-to-end solution architecture, client demos, serverless and AWS EKS deployment in Dev/Prod clusters, CI/CD pipeline for microservices and infrastructure management; perform code reviews, guide technical design, and lead hands-on POCs.
• Collaborate closely with clients on delivery management, roadmap planning, KPI tracking, business use case definition; advise stakeholders on build/buy/rent strategies, LLM and embedding model selection, vector database & observability tools.

AI Architect/Engineer — CGI Technologies, Charlotte NC

June 2023 – Oct 2025

Worked in CGI’s US IP core AI product team- CGI Pulse AI. GenAI services - built with open-source solutions, self hosted as scalable microservices in Azure Kubernetes Service (a) Ask Data - Advanced RAG integration of Onyx -Danswer AI (b) Ask Insight (text to sql) – integration of Vanna AI (c) GenAI inference - self hosted Llama/Mixtral with vllm and Ray serve (d) Monitoring (Prometheus, Grafana, ELK) (e) Responsible AI- Prompt guard

• CGI Agentic AI platform: Built & optimized LangGraph React agents/tools using design patterns (Reflection, Planning, Multi-Agent) for tasks: web search (Serper), productivity (Jira, Office 365), database (Postgres), utilities (REPL, Arxiv, Wikipedia, custom functions).
• Developed Fast APIs for Agent AI config console (no-code UI) and observability using Langfuse; deployed via Helm in Kubernetes. Developed default and custom evaluation metrics (Ragas, G-Eval, Json) by integrating Langfuse APIs with prompt versioning.
• Served as AI Architect for Fannie Mae’s Enterprise Gen AI team (AWS Bedrock/Sagemaker); defined GenAI roadmap, solution vision, milestones across business units. Built reusable accelerators, production-ready AI solutions like chatbots, knowledge assistants.
• Designed enterprise GenAI architecture in AWS with LLM Model Garden layers; recommended LLMs (Llama 3, Mistral, Claude Sonnet), embedding models (Cohere, Titan), vector DBs (OpenSearch, Milvus), benchmark with MMLU, Chatbot Arena, HumanEval
• Built Agentic AI system to automate Foreclosed Property Valuation, integrating appraisal products, Collateral Underwriter, Zillow - reducing valuation approval time by 70% through multi-agent orchestration for document ingestion, validation, red-flag detection.
• Developed multi-agent workflows within AT&T’s Ask ATT platform for customer service automation, including a Team Planner Agent orchestrating sub-agents (Ask Docs API, Ask ATT, Ask Web) with task-specific prompts; Validator Agent for output quality/compliance.

Staff Data Scientist — One Concern, Menlo Park CA

Nov 2021 – May 2023

One Concern is a Data & Insights Risk Analytics Climate Tech startup. As part of Research & Solutions team, we worked with external clients (Insurance, Banking, CRE), Eng, Go to Market, Sales team to build POC’s for clients & productize/scale data science solutions.

• Delivered Resilience & Business Interruption scores for U.S./Japan commercial properties via REST API. These analytics powered 1C-DNA and 1CRX products later integrated into Swiss RE CatNet and Willis Tower Watson platforms through strategic partnerships.
• Data & Analytics product: Develop Exceedance probability/Downtime statistics production pipeline across hazards, return periods, design wireframe; scaled Resilience metrics for 30M properties with Eng team using Kubernetes (GCP), Argo, and CircleCI (CI/CD).
• Mentored data scientists and interns/new hires, led code reviews, and supported pre/post sales client engagements with GTM team (used Tableau).

Data Scientist II — Duke Energy, Charlotte NC

Mar 2019 – Nov 2021

• Built Word2Vec model for failure mode synonyms, used Elastic Search to query work orders, forecast failure mode, detect anomalies
• Provided technical leadership, mentorship to junior data scientists on solutions for water heater failure prediction & sales conversion.
• Developed short/long-term Forecasting models (LSTM, CNN1D, SARIMA) for daily/weekly Disconnect Non Pay work orders using features like seasonality, weather, bill cycles; deployed via batch cron jobs on windows server using CA Workload Automation.
• Developed & deployed a batch Risk Propensity ML model to score customers by likelihood of delinquency. Impact: $5.1M/year
• Developed an unsupervised Gaussian Mixture Model to recommend and rank commercial properties for Energy Efficiency programs in non-native territories; managed and led a team of two data scientists. Impact: $8–10M/year.

Data Scientist II — Brighthouse Financial, Charlotte NC

June 2017 – Jan 2019

• Built Propensity model from diverse data sources to rank & score Financial Advisors most likely to sell Flex & Shield annuity products.
• Built a 3-layer stacked ensemble model (Logistic Regression, XGBoost) across advisor segments and product models; evaluated email campaigns via A/B testing; collaborated with Model Risk/Compliance teams. Impact: $240M in incremental annual revenue ($60M per quarter)
• Led the migration of ML models to SAP cloud in collaboration with architects; deployed Propensity models to dev/prod clusters, automated production workflows (Python), built ML & ETL pipelines (used PySpark), and onboarded the team to the SAP cloud env.

Decision Specialist — R+L Carriers, Ocala FL

March 2013 – June 2017

• Built Time Series Forecasting models (Holt Winter, ETS, ARIMA) using VBA, R to predict key metrics- revenue, shipments, etc
• Automated Forecasting models using SQL/R, reducing run time to 1 hour and saving $800K annually by eliminating manual effort.

Education

MS, Materials Science and Engineering — University of Florida, 2012 (GPA 3.9/4)

BTech, Metallurgical & Materials Engineering — NIT Trichy, India, 2010 (GPA 8.3/10)

Certifications: AWS ML Specialty · AWS Solutions Architect Associate (in progress)

Courses: Machine Learning (Stanford) · Data Science: Data to Insights (MIT)

Publication

• “Effects of cycling on the pseudoelastic properties of CuAlMnNi and TiNi based pseudoelastic alloys”, IJSCS, Vol 1, 2009 Oaktrust tamu library