Looking for a job

Grigoriy Voyakin

@grigoriyvoyakin

Message

Applied AI Engineer at Independent, building agent runtimes and training Gemma 4 12B on runtime trajectories.

Russian Federation

Message

What I'm looking for

I'm open to relocation worldwide and available fully remote.

At Independent, I designed, built and operate Pinocchio Harness, an agent runtime with resumable turns, human approval for risky tools, sandboxed execution and cost telemetry. I also trained and evaluated a QLoRA- and DPO-finetuned 12B model using the runtime’s own trajectories.

I built ShotOps to turn versioned JSON scene specifications into Blender blockouts and video-model renders, with cost-quality benchmarking and a geometric camera audit. My other projects include a video-to-Gaussian-Splat pipeline and a frozen-protocol study of open-vocabulary 3D localization.

At Prequel, I evaluated 20+ image and video generation models and standardized 10+ reusable ComfyUI workflows, reducing setup and iteration time by approximately 30%. Earlier, I integrated a GPT-based agent into Houdini at Assethub Inc and brought a ComfyUI workflow into production at Chemistry Film.

Experience

Work history, roles, and key accomplishments

Current

Applied AI Engineer

Current

Independent

Jan 2026 - Present (8 months)

Designed and built AI systems end to end, including agent runtimes, post-training pipelines, and generative media tools. Developed Pinocchio Harness, ShotOps, and other projects with a focus on production readiness and evaluation.

LangGraph Lang Chain QLoRA DPO Fsdp VLLM Modal Docker Python PyTorch Blender ComfyUI Gaussian Splats RAG MCP fastAPI PostgreSQL SQLite ChromaDB N8n React TypeScript AWS Cloudflare Git CI CD

AI Technical Artist / Generative AI Developer

Prequel

Apr 2025 - Dec 2025 (8 months)

Evaluated 20+ image and video generation models and built 10+ reusable ComfyUI workflows, reducing setup time by ~30%. Integrated detection, segmentation, and pose estimation into generative pipelines, cutting manual steps by ~25%.

ComfyUI LoRAs Image Generation Video Generation Détection Segmentation Pose Estimation Serverless Inference Python NodeJS

AI Engineer / ML Researcher / Pipeline Artist

Assethub Inc

Aug 2024 - Mar 2025 (7 months)

Automated workflows connecting Houdini, Unreal Engine 5, web services, and AI tools, reducing processing time by ~40%. Trained a custom model and integrated a GPT-based agent into Houdini, improving execution stability by ~60%. Built RAG and LLM-assisted workflows for domain-specific search and QA.

Houdini Unreal Engine 5 Python GPT RAG LLM Automation Custom Scripts Server Development

Senior FX Artist / Pipeline Artist / TD

Chemistry Film

May 2023 - Jul 2024 (1 year 2 months)

Put a ComfyUI workflow into production, improving rendering speed by up to 3x on selected tasks and cutting costs by ~20%. Developed custom nodes for studio needs.

ComfyUI Custom Nodes Rendering Optimization Python VFX