ױ Keen - Hey, I build real AI systems that actu...
Hey, I build real AI systems that actually ship.
I work on everything from training and fine-tuning models to stitching together retrieval, agents, and the infra that runs them at scale.
I like getting my hands dirty with data, debugging weird model behaviour, and turning prototypes into stable services.
Main stuff I use day-to-day
* Models & frameworks: PyTorch, Hugging Face Transformers, Diffusers, JAX (when speed matters)
* Model work: fine-tuning, instruction tuning, embeddings, RAG, multimodal pipelines, evaluation and bias checks
* Agents & automation: custom tool-calling agents, orchestration for multi-step workflows, job schedulers and retriers
* Backend: Python (FastAPI), Node.js, gRPC, Postgres, Redis — APIs that survive real traffic
* Serving & infra: Docker, Kubernetes, Triton / ONNX Runtime / vLLM, GPU ops and cost tuning, CI/CD, infra-as-code
* Data & MLOps: data pipelines, validation, MLflow / W&B, monitoring and drift detection
* Frontend / demos: React + Next.js + TypeScript for lightweight UIs and internal tools
If you need someone to untangle a model pipeline, productionize an LLM feature, or just help figure out an inference cost strategy, send me a message.
I prefer rolling up my sleeves and shipping.
I work on everything from training and fine-tuning models to stitching together retrieval, agents, and the infra that runs them at scale.
I like getting my hands dirty with data, debugging weird model behaviour, and turning prototypes into stable services.
Main stuff I use day-to-day
* Models & frameworks: PyTorch, Hugging Face Transformers, Diffusers, JAX (when speed matters)
* Model work: fine-tuning, instruction tuning, embeddings, RAG, multimodal pipelines, evaluation and bias checks
* Agents & automation: custom tool-calling agents, orchestration for multi-step workflows, job schedulers and retriers
* Backend: Python (FastAPI), Node.js, gRPC, Postgres, Redis — APIs that survive real traffic
* Serving & infra: Docker, Kubernetes, Triton / ONNX Runtime / vLLM, GPU ops and cost tuning, CI/CD, infra-as-code
* Data & MLOps: data pipelines, validation, MLflow / W&B, monitoring and drift detection
* Frontend / demos: React + Next.js + TypeScript for lightweight UIs and internal tools
If you need someone to untangle a model pipeline, productionize an LLM feature, or just help figure out an inference cost strategy, send me a message.
I prefer rolling up my sleeves and shipping.
