~/about
❯Name: Vivian Demello
❯Age: ...
❯Occupation: AI Engineer
❯Hobbies: Hackathons, Competitive Programming, Self Hosting
❯CV: resume
~/tech-stack
❯Languages:
❯Libraries & Frameworks:
❯DevOps:
❯Tooling:
❯Languages:
Python
TypeScript
Go
C++
Bash
❯Libraries & Frameworks:
Next.js
React
FastAPI
TailwindCSS
❯DevOps:
AWS
Azure
GCP
Docker
GitHub Actions
Jenkins
Nginx
❯Tooling:
Git
Linux
Neovim
tmux
~/socials
❯GitHub: github.com/vynride
❯X/Twitter: x.com/vynride
❯LinkedIn: linkedin.com/in/viviandemello
❯Email: vynride[at]gmail.com
~/experience
❯Software Engineer Intern · Sozo Labs
Mar 2026 – Present- –Cut LLM input cost ~10× on the highest-volume chat path via Gemini explicit context caching with a static-core / dynamic-tail prompt split, validated by a live API spike serving 99.8% of the prompt from cache
- –Shipped a version-gated home-dashboard A/B experiment (FastAPI + Flutter) to ~2,600 users, lifting trial-to-activation conversion by +8%
- –Classified all 163K users by life-topic with an embedding-based classifier, sharding a ~13h job to ~1h via crc32 partitioning; built a Metabase cohort dashboard surfacing a ~50× conversion gap between engaged and non-engaged users
- –Migrated the platform end-to-end from DigitalOcean to AWS (ECS, EC2, S3, CloudFront) and CI/CD from GitHub Actions to Jenkins, with Bash-scripted VPS deployments
❯DevOps & Backend Intern · AI LifeBOT
Oct 2025 – Mar 2026- –Containerized and deployed 15+ AI and full-stack microservices on Azure Container Apps with Docker and Kubernetes, enabling zero-downtime rollouts
- –Led a zero-downtime migration to a self-hosted VPS (Docker Compose + Nginx Proxy), cutting infrastructure costs by 15%
- –Instrumented Prometheus and Grafana with SLO alerting, sustaining over 99% availability across the fleet
❯AI Engineering Intern · Techmonkey, UAE
Oct 2025 – Dec 2025- –Built a production RAG pipeline (LangChain + ChromaDB, 1.2K+ docs) that lifted answer relevance by 38%
- –Added cross-encoder reranking and HyDE query rewriting to raise top-3 precision from 0.61 to 0.84
- –Streamed responses token-by-token over SSE to cut time-to-first-token by 40%, with a Next.js frontend scoring 98 on Lighthouse
~/projects
❯Redline: Real-Time Voice Crisis-Communication Trainer
- –Turn-based WebSocket loop streaming mic audio through Saaras STT → Sarvam LLM → Bulbul TTS, with live severity & escalation state on the same socket
- –Branching persona behavior driven by JSON-schema structured LLM outputs across 20+ scenarios
- –Coaching pass scores each transcript on five dimensions; shipped on PostgreSQL + Docker Compose + nginx with OAuth
❯RecallAI: AI-Powered Question Paper Analyzer
- –Async multimodal pipeline (PyMuPDF → OCR → Gemini → WeasyPrint) ingesting exam PDFs via FastAPI + Celery
- –Streams live job status to the browser over Redis-backed SSE
- –Per-user AES-encrypted keys behind a JWT proxy + Auth.js OAuth; 500+ users within 4 days of launch
❯NanoCraft: AI-Powered DIY Learning Guide
- –Turns any DIY guide URL into a structured step-by-step visual walkthrough via Gemini structured-output prompts
- –Generates per-step scene images with Cloudflare Workers AI through async background jobs persisted to MongoDB GridFS
- –Context-aware chatbot grounded in scraped project data answers material and step questions
~/achievements
❯Hackathons
- –MLH Vultr Track Winner at Hackprix '26
- –Placed 3rd of 400+ teams at Technex 2025 for building a website-to-RAG chatbot with FireCrawl and Qdrant
- –Reached the semi-finals among 3000+ at the 100xEngineers GenAI Buildathon for Executive AI, an agent that automates Google Workspace and research tasks via Gemini tool-calling
- –Finalist at HackByte 3.0 (Wikipedia Track), IIITDM Jabalpur
- –Finished 5th of 200+ teams at SITnovate 2025 for an AI document summarizer that extracts and condenses key information from lengthy documents and exports to multiple formats
❯Competitive Programming
- –Ranked 3rd at Geek ICPC 2.0, RBU
- –2-Star on CodeChef with a max rating of 1550, having solved 300+ DSA problems
❯Leadership
- –Technical Co-Lead at The CodeBreakers Club, RCOEM, mentoring juniors on DSA, system design, and AI/ML

