Architected AI-assisted tool to accelerate client data loading from source systems to Lyric schema. Used RAG with similar mapping knowledge base to predict column mappings, LLM Agent to generate SQLs for multi-file joins and data translations, and plain English prompts for custom manipulations. All AI suggestions include human-in-the-loop override. Ensured no client data exposure to LLM by operating only on CSV/XLSX headers with guardrails. Implemented evals with LLM-as-judge for testing and prompt injection detection.

Built a Docker-based local development environment using LocalStack for 30+ engineers across 3 teams working on 8 different micro-serviers. Reduced test cycles from hours/days to minutes, delivering 15-20% productivity boost across teams.

Launched a streamlined onboarding questionnaire for client configuration setup, transforming lengthy client interviews into a structured workflow that directly translates answers into Lyric system configuration. Served as architect for cross location teams in NYC and Atlanta.

Launched a comprehensive client management suite handling full client lifecycle from onboarding to billing, mini-app management, and client authorization for Implementation Team. Built micro-front-ends and micro-services architecture enabling multiple teams across NYC and India to work in parallel, delivering a unified experience while centralizing metadata accessed at every login.

Built a tool for bulk employee data imports including work agreements, positions, and earnings/deductions with comprehensive business validation to ensure data integrity for complex HR/payroll setups.

Built event-sourced activity log system with immutable audit trails, capturing every data change with actor identification, before/after values, and business context for auditability, compliance and security.

Identified critical scaling challenge: custom integrations for 30 clients in pipeline would require 25 developers and 6+ months (4 weeks per integration × 4 integrations per client). Built a proof of concept that abstracted integration logic into configurable components. Presented to ADP leadership, secured buy-in from VP of Application Development, hired distributed teams across New York and India, and put together a comprehensive roadmap.

Led, architected and delivered the complete product from 0-to-1 in under 6 months. Personally implemented core algorithms for parallel data fetching and dynamic merging from multiple sources (a lightweight GraphQL-like implementation) using metadata-driven AST parsing and multi-threaded Go routines. Go Text Templating made complex data transformations accessible to non-developers. Established FTP and REST delivery methods to transmit data to integration partners.

Reduced development time for HR, Payroll, Talent, Benefits, and Time integrations from 4 weeks to 4 hours (a 40x improvement), freeing up 20+ developers for other projects. A lean team of 5 implementation associates now manages all integrations, cutting client implementation time by a month. Tool drove wide adoption and earned organization-wide recognition.

Today, the tool generates 30,000 client files per pay period at Lyric. Team in the process of adding Agentic AI features to auto-generate Go text templates from user prompt, further streamlining workflow and reducing cognitive load for Integration Specialists.

Built API support for three distinct access patterns: event-driven delta sync, daily snapshots, and async as-of-date queries, providing flexible integration options for diverse client partner needs.

Clients faced fragmented new hire onboarding across separate HR, payroll, benefits, and time workflows. Assembled a cross-functional SWAT team from internal talent and mentored them on multi-domain HCM knowledge, since this expertise is rare to hire externally.

Architected a unified metadata-driven onboarding flow that dynamically adapts to each client's enabled packages with business validation and bulk import support.

Led distributed team across NYC and South Africa to deliver it in one quarter, turning the client into an early referrable advocate.

Integrated TMBC’s strengths‑based application into Lyric, designing backend calculations to support the approach.

Built performance management apps enabling goal setting, review workflows, and continuous feedback.

Launched Engagement Pulse to measure engagement and alignment, delivering actionable org health insights.

Early member of the Lyric Payroll team. Learned payroll by manually processing real use cases alongside product owners, then built client-facing data entry platforms with bulk import tools and interactive pay data grids. Led a development team in India as technical lead.

Integrated with payroll engines to automate payroll runs, creating Node.js services to translate engine output into Lyric schema.