Rahul Krishnamoorthy
Applied Scientist | Agentic AI | Scalable LLM Systems
- Role
- Data Scientist at ADP
- Location
- Pasadena, CA, US
- LinkedIn followers
- 500 followers
About Rahul Krishnamoorthy
Applied Scientist and AI researcher at ADP with 4+ years of experience building production-grade Agentic AI systems. Spearheading the research and deployment of autonomous agent pipelines for ADP Assist. Currently driving model design, data strategy, and establishing systematic evaluation frameworks. Experienced in agentic routing, context engineering, LLM orchestration, knowledge ingestion, intelligent document processing and deployment of cloud-native microservices.I love connecting with new people, you can reach me at r••••••••@gmail.com
Experience
Data Scientist
Aug 2024 — Present · Pasadena, CA, US
Drove cross-functional collaboration with stakeholders to translate business challenges into agentic AI systems that automate multi-domain HCM workflows through a unified interface.• Architected production-scale agent routing system leveraging vector similarity search, few-shot prompt engineering,and LLM orchestration to intelligently route 320 agents, achieving 95% human-validated accuracy with real-time inference (1.2s latency).• Optimized agent-routing pipeline from 4-stage to 2-stage hybrid architecture by strategically replacing redundant LLM calls with embedding-based retrieval, reducing API costs and latency by 50%.• Implemented a semantic cache by indexing high-frequency utterances and their agent mappings into OpenSearch, further reducing routing latency by 25%.• Engineered the agent knowledge base by prompting LLMs, applying document chunking strategies, and indexing into OpenSearch for the vector similarity search.• Architected entity extraction microservice leveraging LLM function calling with systematic prompt optimization, serving as core NLP infrastructure for ADP Assist platform and enabling multi-step agentic workflows across 60+ production agents at scale.• Designed an intent-drift detection layer within the entity extraction microservice that uses LLM function calling to analyze conversational history, distinguishing between topic persistence and intent shifts across multi-turn interactions to enable accurate downstream agent routing.• Automated end-to-end knowledge ingestion by architecting a service that replaced manual workflows with cron-driven Databricks pipelines, enforced HCM domain metadata standardization via business logic, and persisted versioned artifacts into a DynamoDB store.• Leveraged the versioned DynamoDB store as the canonical data layer for downstream vector indexing and knowledge graph construction, enabling reliable and auditable knowledge retrieval across the ADP Assist platform.
Education
Sri Sivasubramaniya Nadar College Of Engineering
Bachelor of Engineering, Electrical, Electronics and Communications Engineering
2015 — 2019
P. S. Senior Secondary School
High school
2013 — 2015
University of California, Davis
Master of Science - MS, Electrical and computer engineering
2019 — 2021
Find verified contacts for anyone on LinkedIn
Unifers gives sales teams verified emails and direct dials, enriched profiles, and outreach that lands in the inbox.
Free plan included · No credit card required
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.