Wowzer AI is an easy-to-use, safe multi-model AI image website that offers the best image models (Dall-E 3, Stable Diffusion 3, Leonardo).

• Our small engineering team grew Wowzer from 0 to over 100k users and over 2M generated images in 4 months

• Implemented Wowzer backend with TypeScript, Supabase edge functions, Runpod, and PostgreSQL

• Designed and implemented AI safety for Wowzer. User prompts are classified with OpenAI Moderation API and generated images are classified with OpenAI Vision API and Wowzer image classifier

• Designed and implemented image generation worker queue with support for parallel, rate limiting, prioritized image generation with single-GPU servers or serverless API endpoints

• Integrated with various image generation APIs, such as OpenAI Dall-E, Stability AI, and Wowzer APIs at RunPod

• Implemented 18 distinct image generation styles (such as Photograph, Cinematic, Graphic Design, Oil Painting, etc.) by enhancing user prompts, for every image generation model

• Implemented automatic user prompt translation from non-English to English to improve image quality

• Implemented OpenGraph social link sharing

• Implemented a prototype of semantic image search based on prompt vector embeddings

Worked in an analytics team building an analytics data pipeline with TypeScript, GraphQL, Node, Kafka, Redis Streams, Snowflake, and Postgres, on Kubernetes.

Worked on Payment Services Directive 2 (PSD2) integration to Adobe E-Commerce platform.

Worked on Adobe E-Commerce back end (Java Spring Boot, Cassandra) and front end (Vue.js, Node.js).