I build data pipelines that don't break when you add AI to them.
Most Kafka+LLM implementations fail in production. I've spent 5 years with Kafka, Flink, Spark, and real-time systems figuring out why — and how to fix it. Currently at Confluent, building streaming systems that actually work.
Stuff I work with
Work I've Done
Real projects, actual results
Most LLM security tools analyze one message at a time and miss multi-turn attacks entirely. I built StreamGuard — a stateful security layer with parallel detection layers, session history tracking, and honest tradeoffs.
→ Stateless LLM security tools can't detect attacks that span multiple messages — progressive extraction, rephrased blocked attempts, and cross-agent poisoning.
How I built a free platform that spins up personal AI agents in under 2 minutes using Azure Container Apps and BYOK. Spin up, do your work, pause, resume — your context stays intact.
Writing
What I've learned from breaking things in production
4/5/2026
Your DAG completed. No errors. Success metrics green. Then your dashboard showed 75% fewer records than yesterday. Here's what happened — and why it kept happening.
Read more
4/5/2026
Model evaluation: 94% accuracy. Production: wrong predictions everywhere. Your model is fine. Your features are lying to you.
Read more
4/5/2026
Your AI demo was flawless. The model answered every question. Stakeholders approved the budget. You deployed to production. Two weeks later, it's falling apart. Here's why — and it's not the model.
Read more
About Me
I'm Raushan — 5 years building streaming infrastructure (Kafka, Flink, Spark) and AI systems that survive production.
Real-Time Data Systems
Kafka, Flink, and Spark pipelines that process millions of events per day. Stream processing, event-driven architecture, and exactly-once semantics.
AI & RAG Systems
LLM apps with RAG, vector databases, and agents. Built with LangChain, OpenAI, and production-grade observability — not demo-day hacks.
Cloud Infrastructure
AWS deployments with Kafka (MSK), Kinesis, Glue, and SageMaker. Docker, Kubernetes, Terraform — infrastructure that scales without breaking the bank.
Background
Currently a Solution Engineer at Confluent — building real-time streaming systems for enterprise customers. Previously built data infrastructure at startups and enterprises — real-time analytics at 100ms, and ML pipelines at ZS.
Confluent Certified Kafka Developer (CCDAK) + AWS Solutions Architect.
Track Record
30M+
Events/day processed in production
Zero downtime
Full database migration, solo
90%
Reduction in debugging time
40%
Productivity boost via RAG system
Let's talk
Questions about streaming systems, AI, or working together? I usually respond within 24 hours.