Open to opportunities

Hayk Harutyunyan

@haykharutyunyan2

Message

I build production AI and cloud systems—agentic orchestration, Kubernetes optimization, and FinOps cost control.

Armenia

Message

What I'm looking for

I’m looking to build impact at the intersection of agentic AI, distributed systems, and cloud cost optimization—owning end-to-end production architecture, leading teams, and shipping measurable improvements in latency, reliability, and spend.

I build production agentic and cloud systems end-to-end, with a focus on reliable orchestration, measurable performance, and real-world cost control. At Relaunch.ai, I architected and built an agentic orchestration framework (DAG planner, async executor with retry classification and checkpoint resumption, evaluator gates with self-correcting feedback loops, multi-model routing across Claude and Gemini, and credit-based cost controls) powering the product’s autonomous experimentation loop.

I also engineer concrete product outcomes: I built an LLM user behavior prediction engine with >80% correlation to real-world A/B test outcomes, developed a text-to-code agent for autonomous HTML/CSS manipulation that cut experiment design from days to minutes, and shipped a low-latency RAG subsystem and chat interface in React/TypeScript by optimizing retrieval precision. Previously, I led FinOps and Kubernetes workload optimization initiatives at nOps—pivoting toward Karpenter and workload rightsizing to deliver cost transparency and automated, cost-aware control—then at earlier roles focused on scalable AWS automation, event-driven insights, and data ingestion reliability (including a migration to AWS Batch).

Experience

Work history, roles, and key accomplishments

Current

Co-Founder & CTO

Current

Relaunch.ai

Jan 2025 - Present (1 year 4 months)

Architected and built a production agentic orchestration framework (DAG planner/executor, retry classification, checkpoint resumption, and evaluator gates) powering the product’s autonomous experimentation loop. Built an LLM behavior prediction engine with >80% correlation to real-world A/B test outcomes and shipped a text-to-code HTML/CSS agent that reduced experiment design timelines from days t

Orchestration Agentic Workflow (DAG Planner Executor)AI Evaluator Gates Multi Model Routing (Claude Gemini)RAG Async Retries And Checkpointing Credit Based Cost Controls A B Test Outcome Modeling

Lead Software Engineer

Nops

Sep 2023 - Dec 2024 (1 year 3 months)

Led a team of 6 engineers and 2 data scientists to pivot toward Kubernetes workload optimization using Karpenter and workload rightsizing. Delivered cost transparency and automated, cost-aware control of Kubernetes workloads for customers reducing AWS costs by up to 50%.

Kubernetes Optimization Karpenter FinOps Team Leadership

Senior Software Engineer

Nops

Feb 2022 - Sep 2023 (1 year 7 months)

Spearheaded backend engineering for an AWS RI automation tool that provided event-driven, near real-time RI utilization insight. Implemented a migration from large EC2 nodes (Redis + Celery) to AWS Batch, enabling customers to reach >90% RI coverage and improving scalability, reliability, and cost.

AWS Services Event Driven Architecture Redis Celery AWS Batch Scalability Reliability Cost Optimization

Software Engineer

BMAT

Feb 2020 - Feb 2022 (2 years)

Contributed to a data processing pipeline that automated the handling of digital sales reports from music streaming platforms (e.g., Spotify) and generated royalty claim files for music rights holders.

Data Pipelines Digital Sales Report Processing Royalty Claim Generation

Education

Degrees, certifications, and relevant coursework

Middlebury College

Bachelor of Arts

2014 - 2018

Activities and societies: Davis Scholar (academic merit scholarship); Kellogg Fellow (funded 3 months of research in Oxford, England).

Earned a BA with Honors from Middlebury College (2014–2018). Received the Davis Scholar merit scholarship and a Kellogg Fellowship for research in Oxford, England.