Tushar k

At Motorola Solutions, I build production data and observability systems that process 500K+ audit, call-log, and CRD events daily. I’ve improved incident detection by 40%, reduced pipeline failure resolution time by 60%, and shipped an Agentic AI MCP server for faster ticket resolution.

I also automate secure, reliable platform operations through Docker, ELK, OpenTelemetry, Keycloak, Rundeck, and scripted zero-downtime upgrades across 20+ IDAP services. My work has supported 99.5% uptime, reduced infrastructure costs by 25%, and secured access for 200+ users.

Previously at Growleads.io, I built Predicta, an AI analytics platform with 20+ interactive visualizations and natural-language querying using Groq API and Llama 3.3. I developed AI-powered Google Ads intelligence workflows that reduced wasted ad spend by 30% and improved data quality by 45%.

I enjoy building agentic AI systems that connect LLM reasoning with real operational actions. My projects include an autonomous DataOps agent using MCP, LangGraph, FastAPI, Redpanda, Prometheus, Docker, and OpenTelemetry, and I published the tryn-mini7m embedding model on Hugging Face.