Senior Full-Stack
& Cloud Engineer.
Senior full-stack and cloud engineer with 8+ years of experience, currently the sole owner of the AI platform at Nova Dynamic Media, a live-events company. I care about reliability under real pressure over features that only work in a demo, and I've led development teams of 3 to 8 people while staying hands-on with architecture and code rather than stepping back into pure management.
- ~40%
- AWS cost reduction
- 50K+
- concurrent viewers
- 50+
- live productions
Skills & Tools
RAG, Agents & Real-Time Systems
The hardest and most involved work I have built: a production AI platform for live enterprise events — RAG copilots, agentic workflows, and real-time interaction, owned end to end. It is one part of what I do, not the whole of it.
RAG and agentic workflows on AWS Bedrock, with real-time interaction on AppSync. Every answer is grounded in real event data, autonomous agents reason over live metrics and audience questions, and thousands of attendees stay in sync at once. I own it end to end, from the vector knowledge base and the moderator copilot to the summarization and reporting pipeline.
- RAG system on AWS Bedrock grounding answers in event data
- Agentic workflow for querying live metrics in plain language
- QnA grouping and AI report generation, now on Amazon Bedrock AgentCore (building)
- Moderator copilot generating and answering audience questions live
- Live and post-event chatbots plus livestream summarization
- AWS AppSync + GraphQL real-time Q&A, polls, surveys, and quizzes
- Migrated from IoT Core/MQTT to AppSync for better scale
Case Studies
The decisions behind the systems: problem, options, the call I made, and what the numbers did.
- 0double-bookings — enforced in the database, not app code
- 4 providersOpenAI, Anthropic, Gemini, Bedrock — swap by config
- 8 toolsthe agent's entire surface area
Experience
- Sole owner of the AI platform: a production RAG system on AWS Bedrock (S3 Vectors, hierarchical chunking, semantic search) powering a live moderator copilot and post‑event AI assistants with source‑backed answers.
- Own the agentic AI layer end to end — a natural‑language event‑metrics agent (query generation, validation, reasoning steps), automated QnA grouping, and AI report generation — now running on Amazon Bedrock AgentCore with infrastructure provisioned via Terraform.
- Cut throttling failures 60–70% under burst load with cross‑region Bedrock inference failover and exponential backoff, keeping the live copilot responsive at peak.
- Real‑time audience interaction (live Q&A with upvoting and direct replies, speaker teleprompter, polls, surveys, quizzes, synchronized slideshows) built on AWS AppSync + GraphQL + DynamoDB — migrated from the older IoT Core/MQTT setup for better scale. 50+ productions.
- Live AI chatbot for audiences + a post‑event AI chatbot for asking about the presentation, on the same RAG system.
- Livestream summarization and AI‑driven insight reporting.
- Created Mission Control, a unified moderator dashboard.
- Telemetry pipeline: Kinesis Firehose → S3 for storage, Redis/Valkey for live stats; processed with AWS Glue + Athena for reports and AI insights.
- Mass email system: SQS queues import users into the platform, a Step Functions workflow adds them to Mailgun mailing lists (primary provider), with AWS SES as backup.
- Live video transcription and translation with real‑time captions in each viewer's own language.
- Encrypted stored video and reports with AWS KMS.
- QR‑based check‑in app built with Flutter, plus attendance reporting.
- Secure HLS delivery with token‑based encryption/decryption and access control; migrated 1000+ videos to adaptive HLS with AWS MediaConvert, improving quality and reducing bandwidth.
- Cut cloud spend by ~40% via Lambda Graviton migration, EC2 right‑sizing, and DynamoDB capacity tuning.
- Migrated CI/CD from Jenkins to AWS CodeBuild for consistent, multi‑account, cross‑region releases across microservices.
- Co-founded a cosmetics e‑commerce company from the ground up, driving the platform from concept to production.
- Lead a team of 3 engineers across product and infrastructure.
- Set up VPS servers, system infrastructure, and platform architecture.
- Integrated third‑party game provider APIs — session management, webhook callbacks, and real‑time balance updates.
- Built a secure Stripe wallet system in NestJS for deposits, payouts, and reconciliation.
- Bonus and points system based on gaming performance.
- Built internal dashboards for game telemetry, payouts, and incident tracking.
- Second‑hand marketplace with microservices backend and real‑time inventory management.
- Provisioned AWS infra with networking/security and observability; cost optimization built‑in.
- Jenkins pipelines with containerized services for reliable, repeatable deployments.
- Modular CMS powering 12+ client properties (schools, e‑commerce) with reusable components.
- Built Crystal Academy, an online e‑learning platform with courses, tests, and certificates.
- Built e‑commerce platforms including a Taekwondo studio store and Aitken Vanson (shoe retailer) with Stripe‑integrated booking.
- Booking system that increased reservations by 100% with integrated analytics and A/B testing, plus a Google Calendar booking system.
- Stripe payments with high availability; Cloudflare performance optimizations across sites.
- Managed hosting for all client projects on GoDaddy, optimizing frontend delivery with Cloudflare caching, edge configuration, and DNS/CDN routing.
- Agile leadership of 5–8 engineers; introduced CI/CD and code review to reduce regressions.
- ERP with accounting & CRM across multiple departments; role‑based access and audit trails.
- Company registration/renewal workflows and compliance deadline tracking dashboards.
- Search across 1M+ records in 0.3–0.7 seconds using IndexedDB and optimized algorithms.
- Automated reporting cadence (daily/weekly/monthly), saving 15+ engineer hours weekly.
- 5 hybrid mobile apps with offline‑first data sync and conflict resolution.
- Built end‑to‑end logistics features for B2B parcel delivery: order intake, routing, and proof of delivery.
- Geofenced GPS tracking that improved on‑time delivery by 40% with live fleet telemetry.
- Dynamic pricing engine using route length, vehicle type, traffic patterns, and demand.
- Customer service portal with status tracking, SLA alerts, and proactive notifications.
- Delivered travel/tourism websites with custom Laravel modules and WordPress themes.
- Implemented on‑page SEO and performance improvements for better discovery.
- Set up forms, booking flows, and content workflows tailored for agencies.
Education
Projects






AI Event Copilot
I build AI copilots for live enterprise events: RAG-grounded Q&A generation, knowledge-base chatbots, and agents that query event data on demand. I put together an interactive, client-side demo of those patterns (fictional data) so you can try it yourself.
Résumé
Contact
Tell me about your product, timeline, and what success looks like. I respond within 24 to 48 hours.