Ashish Shrivastava

• Designed and engineered production-grade Agentic AI systems by architecting AI/ML systems that combine LLM reasoning, tool use, orchestration, retrieval, memory, and autonomous AI workflows.
• Built enterprise-scale RAG architectures using vector databases, semantic search, knowledge graphs, and contextual retrieval, orchestrated with LangGraph for resilient agent behavior.
• Slashed overall API

• Championed the end-to-end architecture and build of WorkiFi, integrating advanced multi-agent orchestration, hybrid retrieval pipelines, persistent memory, reasoning modules, and monitoring infrastructure; guided team execution to launch in under 9 months, landing the inaugural
paid customer, boosting initial revenue streams by 40%, and scaling multi-domain query handling 3x faster.

• Directed

• Led the design, development, and deployment of a scalable LLM inference pipeline, driving 6.5M+ daily attribute extractions, 65% cost reductions, and 5x faster catalog pipelines through optimized engineering and team coordination.

• Championed partnerships across catalog teams to gather and integrate data, architecting automated enrichment and evaluation workflows that accelerated onboarding,

• Spearheaded the design and rollout of Listnr, a multi-source customer voice analytics system tracking mail, chat, voice, social, surveys, and reviews to surface critical topics and trends; drove cross-team adoption, resulting in 17% higher CSAT, 32% faster resolutions, and 25% uplift in feature engagement through targeted interventions.

• Designed and built the full-cycle development and rollou

• Led the development of a contextualized fallback response system for handling unanswerable dialogue queries using Hugging Face models (with published paper), enabling graceful conversation recovery that boosted user retention by 28%, reduced drop-off rates by 35%, and improved overall engagement metrics by 22%.

• Designed, built, and deployed an advanced question generation system from raw text

• Build a specialized pipeline for attribute and keyword extraction from Intel chip manufacturing documents, delivering precise data structuring that accelerated R&D workflows by 35%, reduced manual annotation efforts by 40%

• Build an advanced extraction pipeline for client data, enabling automated attribute/keyword parsing that drove 35% faster workflows, 40% lower manual efforts.

• Directed academic instruction and team-based learning in AI, Algorithms & Data Structures, and Theory of Computation, guiding 50+ students to achieve higher project success rates, improved problem-solving proficiency

Sep 2017

Sep 2017

Sep 2017

Grade: 9.0

Activities and societies: Research Publication, Machine Learning, Social Network Analysis

Master of Science (By Research) in Computer Science and Engineering

Grade: 69.4

Activities and societies: Computer Science and Engineering, Machine Learning

B.tech in Computer Science and Engineering