I build systems that think and act independently. AWS Solutions Architect · AWS ML Specialty · CKAD → github.com/yashvyas95
I run a fleet of AI agents 24/7 on hardware I own. An orchestrator coordinates ten specialists that keep data fresh, generate insights, research, forecast, write, ship code, and heal themselves when something drifts — on scheduled loops and on demand from my phone. No one babysits them. I built the orchestration and control plane on top of an open agent runtime, and it runs in production against a real data platform, not a demo I wheel out at meetups.
That platform is the point of it all: a self-hosted, multi-source ingestion system — markets, news, weather, research, books — on Spring Boot, React, and Postgres, fully Dockerized. A local LLM turns the raw data into insights and a nightly knowledge graph, all on my own infrastructure.
Before any of this, 5+ years in fintech and enterprise engineering: payment systems, Java microservices, ML pipelines, the usual distributed-systems headaches. Environments where if something breaks at 2AM, your phone rings. That shaped how I build AI systems too — observability, fallback logic, structured decision-making. None of it is optional when your agents run unsupervised.
The thing I keep coming back to is trust. Not "can an agent do something smart once" but "can it keep doing the right thing when nobody's looking." That's a harder problem than most people realize — and it's mostly an infrastructure problem, not a model problem.
Contributing to open source when I can. Always happy to talk with anyone working on agent systems or applied AI in regulated industries.
Languages
AI / ML
Backend
Frontend
Cloud & Infrastructure
Data & Messaging
Observability & CI/CD
| Project | Status | What it is |
|---|---|---|
| Multi-Agent Infrastructure | running |
Autonomous agents with memory, tool execution, and cross-agent coordination on dedicated hardware |
| Agent Dashboard | running |
Central UI for monitoring agent sessions, knowledge graphs, timelines, and ops |
| Custom Agent Runtime | running |
Built from scratch — multi-model routing, Telegram/Gmail integration, structured fallback logic |
| Open Source Contributions | active |
| Certification | Provider |
|---|---|
| AWS Machine Learning Specialty | Amazon Web Services |
| AWS Solutions Architect Associate | Amazon Web Services |
| Certified Kubernetes Application Developer (CKAD) | CNCF |
- Design & Architecture Pattern Reference Guide — CQRS, event sourcing, microservice patterns, and when to actually use them
- AWS Machine Learning Specialty Prep Guide — End-to-end study guide covering SageMaker, data engineering, and ML ops
- Dynamic Programming — Deep Dive — Pattern-based breakdown of DP problems with intuition-first explanations
Books that actually shaped how I build things:
| Book | Why it matters to me |
|---|---|
| Designing Data-Intensive Applications — Martin Kleppmann | Changed how I think about distributed state. I reference this constantly when designing agent memory and coordination. |
| Deep Work — Cal Newport | The reason I can context-switch between enterprise code and agent infrastructure without losing depth. |
| Design Patterns — Gang of Four | Less about memorizing patterns, more about having a shared vocabulary when designing systems with other engineers. |
| Effective Java — Joshua Bloch | Still the best book on writing code that doesn't bite you six months later. Directly applicable to the Spring Boot services I've built. |
Building autonomous systems that earn trust. Open to conversations about agent infrastructure, applied AI, and distributed systems.