Get a reply from this employer
Your resume and cover letter, tailored to exactly what they’re hiring for.
Pengiklan Anonim seeks a capable Backend Engineer to own the real-time streaming backend for an AI avatar platform. You will drive the lipsync and voice-cloning pipelines, coordinate LLM and TTS/voice services, and deploy on GCP with GPU workloads.
You will excel in asynchronous Python, fast APIs, and scalable microservices, delivering low-latency, cost-aware solutions, and collaborating with frontend and infra teams to ensure robust state management.
We are seeking an autonomous, highly skilled Backend Engineer with approximately 5 years of professional experience to drive our real-time AI avatar platform's backend. You'll own the systems that turn a user's voice into a GPU-rendered, lip-synced avatar video streamed back in real time — spanning the lipsync inference pipeline, the conversational orchestration server, and the voice-cloning/TTS service, plus their deployment across GCP.
You are a versatile backend engineer who thrives on asynchronous Python, real-time protocols, and GPU-backed inference services. You understand both the low-latency demands of streaming systems and the operational realities of running GPU workloads in production, allowing you to design services that are fast, resilient, and cost-aware.
Develop and scale the WebSocket/Protobuf streaming layer that carries audio in and lip-synced video out with tight latency budgets.
Extend the orchestration server (FastAPI, asyncio) that coordinates LLM, TTS/voice-cloning, and lipsync services per session.
Work with PyTorch/CUDA-based lipsync and voice-conversion models — pipeline allocation, warm pools, batching, and multi-user concurrency on shared GPUs.
Design and build a CPU-only rendering path for realistic avatar output, so avatar sessions can run without dedicated GPU inference where full lipsync fidelity isn't required — reducing per-session cost and unlocking scale-to-zero / low-cost deployment tiers
Build and evolve REST and WebSocket APIs, and Protocol Buffer message contracts shared across services and the frontend client.
Manage Docker-based deployments to GCP Cloud Run/GCE and GPU cloud providers (RunPod, Vast.ai), including config promotion, health/readiness checks, and cold-start/scaling tradeoffs.
Partner closely with the frontend engineer to co-design WebSocket/REST contracts and data flow between client and backend services.
Work with Redis for session/state management and GCS/Vertex AI for storage and RAG-backed retrieval.
Work directly on live, production-scale AI systems reshaping the retail industry.
Lead front-end development with high autonomy, working within a fast-moving team that respects technical excellence.
Join a collaborative environment where speed, creativity, and clean code are highly celebrated.