Vikas Agarwal

I’m an AI Engineer with 8+ years building production AI systems and data platforms, with a hands-on focus on LLM application design, RAG, AI Agents, and retrieval quality. I care about evaluation rigor—using structured quality metrics, golden datasets, and regression testing—so grounded outputs and silent-failure prevention are built into the pipeline, not added later.

At Flick2Know Technologies (FieldAssist), I lead AI and data engineering and own architecture from RAG pipelines to ClickHouse warehousing. I’ve scaled core systems to 100K+ users and 1.5M+ daily transactions at 99% SLA, driving sharding, partitioning, federation, and geo-redundant replication, while mentoring an 8-member team and delivering end-to-end with zero-downtime releases.

My standout work includes an agent-driven Universal Data Integration Platform that processes 50M+ records daily in production and runs 500+ workflows at 150+ executions/sec—replacing a costly vendor workflow with an in-house platform that delivered 500+ integrations in a year. I also built a Self-Analytics Portal with star-schema + RAG, establishing a retrieval evaluation harness (context precision/recall, faithfulness) to prevent regressions and cutting reporting cycles from 3–4 days to 1–2 hours, plus production anomaly detection and an agentic code migration workflow for .NET upgrades.