Mustafa Ayyub

Developing Python-based PoC prototypes with REST APIs to automate complex data retrieval processes Built Multimodal RAG pipeline using ColPali for visual document retrieval, eliminating OCR errors Integrated Llama Vision models via Groq API achieving sub-second inference latency Engineered Visual Embedding search system with logging and monitoring for charts and tables interpretation Participating in team meetings using Jira for task tracking to align prototypes with requirements

Led DocuTalk-AI: RAG-based assistant using LangChain, MiniLM, Pinecone, OpenAI. Built Streamlit + Python full-stack system for 500+ users. Integrated Stripe (tiered billing) and Firebase Auth. Achieved 85%+ query accuracy on 1K+ interactions. Sub-200ms responses via semantic search & caching. Managed 2-person Agile team with sprints & reviews. Built fault-tolerant pipelines with logging & retries.

Engineered APIs & automation systems, reducing processing time by 30% Built cron jobs in Next.js handling 10K+ records/day with zero downtime Integrated HubSpot with 100% schema compliance & real-time sync Reduced API latency via caching, Redis, and optimized calls Handled 500K+ monthly API requests with 99.9% uptime Created Grafana dashboards & CI/CD with GitHub Actions Tech: Next.js, Node.js, Redis, PostgreSQL, Docker, HubSpot, Zenoti

Developed 15+ RESTful APIs in Ruby on Rails for fintech/comms platforms with 99.8% uptime Optimized PostgreSQL queries (800ms → 200ms) and designed scalable DB schemas Integrated Stripe & Twilio; documented APIs with Swagger (50% faster client onboarding) Achieved 95% test coverage using RSpec & Jest Participated in Agile ceremonies; contributed to scalable system architecture Tech: Ruby on Rails, PostgreSQL, Stripe, Twilio, Swagger, RSpec, Git