Hayk S.

I'm a senior full-stack developer with over 8 years of experience delivering scalable, high-performance web applications for startups and global companies. My career has taken me through a range of industries—from healthtech and fintech to gaming and e-commerce—where I've helped teams turn complex ideas into elegant, efficient digital products.

My technical strengths lie in React, Next.js, TypeScript, Node.js, Python and Mobile Development with React Native/Flutter with deep expertise in building intuitive frontends, robust APIs, and cloud-native backends. I’m also experienced in blockchain integrations, smart contract interactions, and NFT payment systems. Whether it’s creating custom UI components, optimizing PostgreSQL queries, or developing serverless functions on AWS, I bring a versatile, detail-oriented mindset to every project.

What is more, I also possess Agentic AI & LLM Systems integrations in my experience.

• Designed and deployed LLM-based systems using GGUF models and text-generation-webui in Dockerized environments (RunPod)
• Built backend services enabling AI agents to interact with external tools and APIs
• Implemented stateful workflows for dynamic user interactions (subscription systems, multi-step processes)
• Developed scalable infrastructure for hosting and managing multiple AI model instances
• Integrated AI-driven logic into real-world applications (healthcare dashboards, automation tools)

Beyond client work, I’ve contributed to internal tooling and performance tuning initiatives that significantly improved app load times and maintainability. I’m passionate about clean architecture, DX (developer experience), and building interfaces that feel as good as they look.

I value collaboration, transparency, and continuous learning. My mission is simple: build meaningful software that solves real problems, scales beautifully, and creates great user experiences.

I’ve built and orchestrated multi-agent workflows where different components handle planning, execution, and response generation. As part of this:

• I implemented an agent harness responsible for coordinating multiple steps (task decomposition → tool execution → response synthesis)
• Used graph-based orchestration patterns (via LangGraph) to manage state transitions, branching logic, and execution flow between agents/tools
• Designed stateful agents with checkpointing, allowing workflows to persist intermediate steps and resume execution when needed
• Built evaluation layers to validate outputs before triggering downstream actions (e.g., ensuring structured responses and consistency)
• Integrated tool/function calling, enabling agents to interact with backend APIs and external services

A concrete example is an AI-driven mobile assistant concept I’ve been working on, where the agent:

• Interprets high-level user intent (e.g., planning tasks or routines)
• Breaks it into actionable steps
• Dynamically calls tools (APIs, data sources)
• Maintains memory across steps and sessions
• Iteratively refines outputs before presenting results

This system required orchestration, memory management, checkpointing, and evaluation, all of which align closely with production agentic AI patterns.

Overall, my focus has been on building reliable, backend-driven agent systems rather than just prototyping, so I’m comfortable designing and scaling these architectures in real-world environments.