2024 — 25·Backend Developer · DevOps
Aspirez
AI-powered SaaS for personality insights. Architected the Go backend, real-time chat over WebSocket + Redis, and OpenAI/Deepgram integrations for LLM chat, TTS, and STT. Onboarded 1,000+ users in two months and automated bulk email to 50k via AWS SES.
› Overview
A freelance SaaS engagement I picked up alongside my full-time role at Buymed. Aspirez uses LLMs to surface personality insights via the RIASEC framework — chat with an AI mentor, take guided assessments, and get back a personal report.
› The problem
The team wanted to launch a marketing campaign on a tight timeline. We needed a backend that could handle real-time AI chat, voice in and out, bulk emails, and a steady onboarding flow — without standing up dedicated DevOps.
› What I built
- Architected a Go backend with REST APIs and a WebSocket layer for real-time chat.
- Integrated OpenAI for LLM responses and Deepgram for TTS + STT — wiring it so users could speak to the assistant and get spoken replies.
- Designed a Redis-backed conversation cache so chat sessions stayed warm and the LLM responded fast.
- Set up GitHub Actions CI/CD for rapid feature releases — Google/Apple auth, S3 uploads, and report generation went out within weeks.
- Wrote the bulk-email pipeline on AWS SES with retry, batching, and unsubscribe handling.
› Outcomes
- 1,000+Users onboarded in two months via the campaign
- 50,000Bulk emails automated through AWS SES
- 80%Manual email processing time eliminated
- 25%Backend scalability improvement via Redis caching