saif hany

• Designed and built a Retrieval-Augmented Generation (RAG) chat pipeline for a multi-platform social media analytics system
(Twitter/X, TikTok, Instagram), enabling natural-language querying over indexed reports.
• Implemented hybrid Elasticsearch retrieval combining dense KNN vector search (BGE-M3 embeddings) with BM25 keyword search,
fused via Reciprocal Rank Fusion (RRF) to improve search relevance and ranking accuracy.
• Developed an LLM-based semantic router to dynamically select the correct Elasticsearch index and retrieval strategy per query,
replacing static routing logic with context-aware, platform-specific decisions.
• Built platform-specific query builders and user-target filtering logic to handle cross-platform data heterogeneity across seven distinct
report types.
• Worked across a two-service FastAPI/SQS architecture, integrating OpenRouter-hosted LLMs, PostgreSQL, and Elasticsearch to
deliver a production-grade AI analytics assistant.

• Led a major backend revamp for Buffalo Burgers customer app, website, and call-center admin panel, supporting high traffic from
140+ branches across Egypt.
• Resolved severe order-history performance issues by introducing an archived database for legacy orders and separating archived vs.
active queries; applied combined indexing to speed up admin search.
• Migrated the order module into a standalone microservice using RabbitMQ (Pub/Sub), and added Redis caching for static data and
refresh tokens to reduce database load.
• Built a reusable NPM package to unify order calculations across mobile and web, preventing client-side pricing discrepancies.

• Developed serverless backend services (Node.js, TypeScript, Firebase Firestore) and core modules — authentication, orders, wallet,
products, cart, reviews — with third-party integrations for wallet, notifications, and shipping.

• Built Python scripts to migrate and reindex large historical datasets from MongoDB to Elasticsearch using bulk indexing to prevent
timeouts.
• Designed a continuous indexing strategy for incoming data, leveraging Elasticsearch language analyzers to improve search accuracy
and relevance.

• Built backend services for a fintech trading SaaS, integrating real-time data providers (Twelve Data, TradingView, MetaTrader);
optimized for high-volume transactions and hardened security.
• Developed and maintained real-time market-data pipelines and trading APIs, ensuring low-latency delivery of price feeds and reliable
order/transaction handling under high concurrency.

• Developed a scalable REST API (Node.js, Express) applying SOLID principles; added Redis caching and a MariaDB request-logging
system for performance and reliability.