Rajakumar Sundaramoorthy

Embedded Engineering Leadership

Lead a 7-person team at Accenture embedded with a top-tier wealth management client (15 months, ongoing). Working alongside the client's product, engineering, compliance, and executive stakeholders to ship AI capabilities into production workflows. Own hiring, coaching, technical discovery, delivery cadences, and translation between engineering and business. Hands-on in the codebase and on architecture throughout. This is the shape of work applied AI and vertical AI product teams do: product engineering tightly coupled to customer workflows.

Customer Partnership & Roadmap Shaping

Designed and shipped tailored AI capabilities into the client's record-keeping system: context engineering with Kiro and agentic developer tooling driving a 22% reduction in code review turnaround; utility AI agents recovering ~1 hour per employee per day. Partner directly with the client's product, engineering, and compliance teams to shape what gets built next: turning customer learnings into engineering priorities, joining technical conversations with their security and audit reviewers, owning the platform-readiness work each launch requires.

Agentic Systems Engineering

Designed and shipped a multi-agent due diligence runtime with parallel agent orchestration, tool calling, structured outputs, ground-truth evals with fuzzy matching, and SQLite tracing of every LLM and tool call. Built for provenance, auditability, and citation-grade output: the pattern high-trust enterprise customers need before adopting agentic workflows. The eval framework also gives reliable signal on output quality across domain use cases, the kind of evals work that informs both product and research. Companion projects: multi-tenant RAG platform with ACL-safe retrieval and integrated eval layer; Financial Journey Intelligence Agent generating next-best-action recommendations for wealth advisors. All on GitHub.

0→1 Team Building

Founded and led two engineering teams at The Hartford from zero (12 engineers across data preparation and entity resolution, three years, no inherited playbook). Defined mission, hiring bar, technical bar, and operating cadences. Coached an engineer hired straight out of college into a Product Owner role through structured 1:1s and stretch assignments.

Regulated FS & Insurance Domain Depth

20 years inside the regulated FS verticals frontier AI companies increasingly serve: insurance (Hartford), retirement and asset management (TIAA), wealth management (Accenture engagement with a top-tier wealth manager). Working knowledge of insurance claims and fraud analytics, entity resolution and 360° business graphs for cross-sell and underwriting, retirement income products with daily reconciliation requirements, advisor next-best-action workflows. Plus the operational layer enterprise customers expect: procurement, security review, compliance attestation, audit posture, multi-tenant isolation requirements.

Multi-agent orchestration, tool calling, structured outputs, production RAG pipelines, evaluation frameworks, MCP integration, prompt engineering and context-engineering tradeoffs, agent skills and sub-agents.

Technical discovery and scoping, customer architecture review, integration patterns (REST, SSO/SAML/OAuth), white-glove enterprise deployments, post-deployment adoption measurement, repeatable starter repos and integration templates.

AWS (Solutions Architect Professional), Python, event-driven architectures, Kafka, Snowflake, Databricks, Spark, distributed systems, observability tooling.

Audit-grade compliance, incident response, on-call operations, deploy safety, business-signal detection, end-to-end monitoring (Elastic, Splunk, Grafana).

Accenture, Charlotte, NC  |  Retirement & Wealth Management Client Engagement

Lead a 7-person delivery team (3 engineers, 3 business/technical analysts, 1 Quality Engineer) embedded with a top-tier retirement and wealth management client. Own hiring, coaching, technical direction, sprint cadences, and executive stakeholder partnership across multi-workstream production engagements (15 months, ongoing).

• Drove adoption of LLM-assisted developer tooling and agentic AI capabilities, driving a 22% reduction in code review turnaround, accelerating sprint velocity.
• Designed and shipped utility AI agents recovering ~1 hour of capacity per employee per day across business and technical functions.
• Defined secure, auditable integration blueprints connecting external AI platforms with internal systems, satisfying compliance, governance, and responsible AI requirements at scale.
• Evangelized context engineering and AI-assisted development with Kiro across the engagement, embedding modern AI development practices into the client's record-keeping system workflows.
• Built and delivered a guaranteed annuity product end-to-end: partnered with the client's product, engineering, and compliance teams from technical scoping through production launch.
• Integrated Capitalize (third-party retirement rollover platform) with the client's existing product portfolio, enabling streamlined participant rollovers across their record-keeping ecosystem.

TIAA, Charlotte, NC

• Designed the foundational data layer for TIAA's Secure Income Account, a novel financial product requiring reliability across daily ledgering, reconciliation, and reporting.
• Architected an end-to-end observability system surfacing business and system signals across data pipelines, reducing mean time to detection by hours and measurably improving operational trust.
• Shipped cloud-native data pipelines (AWS) powering analytics, audit reporting, and operational dashboards.
• Built a Configuration-as-Code framework for synthetic and scenario-driven test data generation, cutting validation time for new features by ~40%.

The Hartford Insurance, Charlotte, NC

Founded and led 2 data engineering teams from scratch (12 engineers across data preparation and entity resolution) over three years. No inherited playbook. Owned hiring, coaching, technical direction, operating cadences, and stakeholder partnership with marketing, cross-sell, and underwriting leadership. Mentored an engineer hired straight out of college over three years to a Product Owner role.

• Data Preparation team: Sourced structured and semi-structured data from multiple systems and conformed it to a common model, powering downstream ML, analytics, and decisioning across the enterprise.
• Entity Resolution / ERaaS: Built Entity Resolution as a Service, a self-service identity resolution platform creating 360° business entities used by cross-sell, marketing, and underwriting teams for risk and pricing decisions.
• $6.2M/year in savings: Architected and shipped a Claims Analytics System that improved referral quality for fraud investigations: directly measurable adoption-driven impact.
• Re-architected the Commercial Market Reference Asset as a cloud-native AWS platform (S3, EMR, Glue), improving data lineage, scalability, and regulatory compliance.

Multi-Agent Due Diligence Analyst

Multi-agent due diligence pipeline that researches companies in parallel via tool calling and structured outputs, producing a canonical JSON report with per-claim confidence, inferred source tier, ground-truth evals with fuzzy matching, and full cost/token observability (SQLite tracing of every LLM and tool call, Streamlit replay). Markdown and PDF derive from the JSON. Built for provenance, auditability, and citation-grade output: the pattern legal, asset management, and compliance customers actually buy.

Enterprise RAG System

Production-grade RAG platform with event-driven ingestion, tenant isolation, ACL-safe retrieval, and grounded LLM answers with citations. Combines lexical and vector search with an integrated evaluation layer, observability pipeline, and Docker-based local development.

Financial Journey Intelligence Agent

Real-time streaming system maintaining per-customer journey state, firing an LLM agent only on meaningful behavioral triggers (~7% of events) to generate next-best-action recommendations for wealth advisors. Built with a 5-stage intent-scoring state machine, RAG-retrieved product knowledge, and synthetic data generators, architected to map directly to Kafka in production.