AI integration
LLM agents, RAG pipelines, tool use and MCP servers — wired into real production systems, not demos. Recent work: 40% lower inference costs, sub-second retrieval over 500K+ embeddings, document pipelines cut from ~8s to under 2s.
AI Engineering · Business Analysis · Delivery Management
Senior engineer and delivery lead for teams putting AI into production.
Andrew Motsyk. 13+ years across analysis, architecture, development and release — for Nike, Mars, GAP and Wix. Now building production LLM agent systems and RAG pipelines, plus the discovery and analytics work that decides what is worth building in the first place.
13+ years · engineering, analysis & delivery5+ years · as technical project managerNike · Mars · GAP · Wix · shipped forRemote, worldwide · any timezone · PSM certified
Selected results
40%lower LLM inference costs
500K+embeddings, sub-second retrieval
$1M+secured in contract extensions
40M+users on the network I replatformed
About
I'm a senior software engineer and manager with 13+ years across analysis, design, development, release and architecture — lately focused on integrating AI-powered workflows into production systems: LLM agents, RAG pipelines on Qdrant, tool use, MCP servers and cost-aware prompt engineering.
I also run delivery, not just the build. Nearly six years as Technical Project Manager at Grid Dynamics: leading cross-functional, multinational teams from pre-sales and estimation through discovery, requirements and backlog ownership to roadmap, release planning and hand-off — and I hold product ownership on the AI platform I'm building now. PSM certified.
Outside of software I have vast experience in live, studio and programmatic audio, plus video recording, mixing and editing — useful more often than you'd expect on media-heavy products.
What I'm hired for
AI is the through-line — from working out where it actually pays off, through building it, to owning the delivery around it.
Business analysis & analytics
Discovery, requirements, and the numbers that check them. Stakeholder workshops become a backlog with acceptance criteria a team can actually build from; product metrics and A/B tests close the loop afterwards. At Wix/DeviantArt that caught a roadmap flaw worth an extra 25% of engaged audience post-launch. Includes AI-readiness assessment — which parts of a process an LLM genuinely improves, and which it doesn't.
Delivery & project management
Owning delivery end to end: scope and estimation, roadmap, release planning, dependency and risk management, and the client relationship around all of it. Nearly six years as Technical Project Manager at Grid Dynamics delivering for Nike, Mars and GAP — development time down 20%, build routines 2–3x faster, and $1M+ secured in contract extensions.
Web & product development
End-to-end delivery of web products: architecture, React/TypeScript front-ends, Node.js back-ends, releases and everything between. The same practice that shipped for Nike, Mars, GAP and Wix — 30%+ faster load times, 2–3x faster builds.
Consulting & technical leadership
Brought in as an outside pair of eyes: architecture and codebase reviews, project rescue, pre-sales support, and building the team that takes it forward — hiring, mentoring and the working practices that stick after I leave.
Delivered for
Nike
Mars
GAP
Wix / DeviantArt
Experience
Oct 2025 — present
Staff Full Stack AI Developer / Product Owner
NDA
Designed and built a modular AI processing platform: ingestion of documents, images and audio, vector indexing on Qdrant, LLM extraction and multi-step agent workflows (Anthropic SDK), GPU-accelerated processing. Cut LLM inference costs 40%, achieved sub-second retrieval over 500K+ embeddings, and brought document pipeline latency from ~8s to under 2s.
Jan 2020 — Oct 2025
Lead Developer / Technical Project Manager
Grid Dynamics
Delivered complex projects for Nike, Mars and GAP; secured $1M+ in contract extensions. Re-architected front-end and APIs for 30%+ faster load and response times while cutting development time 20%, and made build routines 2–3x faster through automation.
Dec 2017 — Jan 2019
Lead Full-Stack JS Developer
Wix / DeviantArt
Replatformed one of the oldest social networks with 40M+ active users. Fixed a roadmap flaw that engaged an additional 25% of the audience post-launch, and introduced a parallel design/dev workflow saving ~30% of development time.
Jun 2016 — Oct 2017
Full Stack Web Developer
Leantegra
Built a high-load real-time location system (RTLS) in the IoT domain on a broker-topology microservice architecture. In the industry since May 2013.
Stack
AI
- LLM agent systems
- RAG / Qdrant vector search
- Tool use & function calling
- MCP server development
- Cost-aware prompt engineering
Analysis & product
- Requirements & backlog ownership
- Stakeholder workshops & interviews
- Pre-sales, discovery & estimation
- A/B testing & experiment design
- GA4 & product metrics
Front-end
- TypeScript / JavaScript
- React, Next.js
- Redux, MobX
- SVG, Canvas, WebGL
- Animations, a11y, performance
Back-end
- Node.js, Express, NestJS
- WebSockets / SSE, Redis
- AWS, microservices
- MongoDB, Neo4j, Postgres
- Python
Tools & QA
- Docker, Kubernetes
- Jest, Cypress, Playwright
- Jenkins, Git, Bash
- Storybook, Figma
- Video & audio streaming
Delivery & leadership
- Delivery ownership, end to end
- Roadmap & release planning
- Estimation, scoping & risk
- Cross-functional multinational teams
- Hiring, mentoring & team building
- PSM certified, Agile / Scrum
Background
Education
- B.Sc. Computer Science · Donbas State Machine Building Academy, Kramatorsk (2003 — 2007)
- Master of Law · National Law Academy of Ukraine named after Yaroslav the Wise, Kharkiv (2007 — 2012)
Languages
- English — C1+
- Ukrainian — native
- Russian — native
Contact
Open to remote engagements — development, analysis, delivery, or all three. Any timezone.