I Build Proactive Agentic AI Systems That Replace Operational Roles
I design context-aware, production-grade AI agents that autonomously run finance ops, support triage, and DevOps workflows—with human-in-the-loop safety, auditability, and measurable business impact.
These systems don't just assist users. They own workflows end-to-end, reason over real-world context, and act reliably in production.
How I'm Different
Most AI projects are chat interfaces. My systems are autonomous agents that own real work.
Proactive
Agents act without prompts. They monitor, reason, and execute based on triggers and schedules.
Context-Aware
Agents reason over history, state, and relationships—not just the current prompt.
Agentic
Plan → Decide → Execute → Learn. Agents own workflows end-to-end.
Safe
Human-in-the-loop gates for risk. Autonomy has boundaries.
Auditable
Every decision explainable. Full trace from action to outcome.
Production-Grade
LangGraph state machines, typed agents, observability, and cost controls built-in.
Core Architecture Patterns
LangGraph state machines, PostgreSQL + pgvector, Redis queues, Neo4j knowledge graphs, Pydantic-AI, and production observability.
GitHub Projects
Open source work and production code
YouTube
Tech talks, tutorials, and agentic AI systems
Featured Videos
Watch my latest tech talks and tutorials on AI systems
Open in YouTubeTechnical Writing
Thoughts on agentic AI systems and production engineering
Transformers: The Architectural Backbone of Modern AI
The Death of Recurrence and the Birth of Parallelism 🚀 • Before 2017, sequence modeling was dominated by Recurrent Neural Networks (RNNs) and LSTMs that processed data word by wo...
Stop Serializing Your AI Agents 137x Speedup with Parallel Execution
Many AI agents currently in production underperform because they are built as simple linear chains where tasks execute one after another. When every reasoning step tool call and A...
Why AI Apps Fail: The HLD/LLD/DSA Framework (Save $190k/Year)
The hard reality of the current AI landscape is a massive gap between experimental potential and operational reality. Industry data suggests that between 80 percent and 95 percent...
The Trinity Stack Building the zero dollar Private AI Employee
The future of the autonomous workforce is no longer a corporate secret locked behind expensive cloud subscriptions. By combining three cutting edge open weight models into the Tri...
ElevenLabs: $99/mo vs. Kokoro + VoxCPM: $0 (Better Quality) ️
https://medium.com/media/b88c597452be728f944b687fccc5406e/hrefFor years, high-quality voice synthesis was locked behind expensive SaaS paywalls, with content creators often paying...
Datadog Steals $10k/mo. SigNoz Does It Free.
For many high-growth engineering teams, Datadog’s bill often feels like maintaining a luxury yacht, with monthly costs easily hitting $10,000 as architectures scale. While Datadog...
I don't build AI features. I build reliable autonomous systems that take responsibility for real work.
If your team is drowning in operational overhead, I can help you design an AI system that actually removes it.