Building LLM products and backend systems for creator monetization, automated conversations, upsells, tipping, and operational control.

• Designed production infrastructure for AI agents, orchestration, tool use, retrieval, long-term memory, and reliable execution.
• Built systems that automate creator conversations, improve monetization workflows, and support platform-scale AI operations.
• Implemented classification, moderation, and guardrail systems for safer and more consistent AI behavior.

Led production LLM feature work across Replika's large-scale consumer AI companion experience for more than 38 million users.

• Owned technical direction and delivery for LLM-powered product work across chat, companion behavior, and backend reliability.
• Improved conversation quality and reliability for next-generation AI companion systems.

Built production backend and LLM systems for companion features, multimodal AI workflows, and a fast-growing AI chat product.

• Developed multi-agent systems with Python, FastAPI, Redis, RAG, and LLM infrastructure for millions of users.
• Designed backend and LLM integrations for a new AI chat and roleplay app built by a two-person team.
• Shipped text, image generation, image recognition, extraction, formatting, and moderation use cases.

Led backend delivery for an AI fashion platform across product features, architecture decisions, and AI image-generation systems.

• Guided backend execution across Python, Golang, and TypeScript services.
• Designed a multi-provider image-generation backend combining internal GPU-hosted models with external APIs.
• Owned backend development for an AI fashion assistant that helped users generate and iterate on clothing designs.

Built and scaled core backend systems for an AI fashion platform, from image processing and service architecture to payments.

• Cut processing costs by 40% and improved speed by more than 50% by refactoring Python image-processing services with threading and multiprocessing.
• Rewrote core services from Django to a Socket.IO and microservices architecture using Flask, Redis, MongoDB, and GCP Pub/Sub.
• Rebuilt Stripe webhooks and payment flows, reducing payment failures from 80% to 0%.

Developed backend features, integrations, and data-processing workflows for analytics and client-delivery systems.

• Built platform features with Flask, JavaScript, SQLAlchemy, and integrations with Google, Facebook, and Supermetrics APIs.
• Accelerated client delivery from W-1 to D-1 with S3 retrieval and ETL processing pipelines.
• Reduced team workload by 50% and cloud costs by 30% by distributing data-processing scripts across virtual machines.

Built API connectors, ingestion workflows, and internal backend tooling.

• Developed Python connectors for large-scale JSON ingestion, automated S3-to-Kafka workflows, and documented APIs and team processes.