Muhammad Farhan

• Consolidated 15 chained microservices into a single TypeScript state machine on Cloudflare Workers, cutting end-to-end latency from 3–8s to 1–2s and making AI-assisted legal workflows feel real-time.

• Designed a Python intent classification router with Redis caching to gate expensive LLM calls, cutting API latency by 35% and per-session compute cost by 70%.

• Designed multi-tenant REST APIs with RBAC and Stripe-integrated tiered billing, enabling the platform's first paid subscription revenue across 3 pricing tiers.

• Built WebSocket and streaming layers for a real-time voice assistant, increasing average session length by 15%.

• Engineered a Python REST API voucher distribution system handling millions of daily financial transactions, doubling monthly voucher revenue and activations.

• Delivered a Java with PostgreSQL backend for a distributed POS system across 5 teams, increasing peak-load throughput by 29% and reducing checkout failures by 40% during peak hours.

• Led microservices decomposition with Docker and API Gateway on AWS CI/CD, accelerating release cycles by 40% and cutting time-to-market for new product launches.

• Rolled out Prometheus and Grafana observability templates across 10+ services, reducing user-facing complaints by 52% through proactive degradation detection.

• Built a config-driven Go backend for A/B experiment management, cutting deployment time by 50%.

• Optimized large-dataset retrieval across 100M+ user profiles via FAISS vector search and Cassandra query tuning, lifting recommendation CTR by 3%, equivalent to millions of incremental clicks per month.

• Built a real-time ML feedback pipeline on Kafka, delivering sub-minute model signal updates to recommendation systems and improving recommendation relevance for 100M+ monthly users.