Talha Niazi

I've built production AI systems across Turing, Buildables, and NACTA, from adversarial LLM evaluation suites to automated NLP ingestion pipelines. My work focuses on reliable reasoning, tool use, schema compliance, retrieval accuracy, and safety boundaries.

At Buildables, I lead technical roadmaps and architecture for machine learning and generative AI applications, while mentoring engineering tracks on scalable NLP workflows, transformer fine-tuning, and structured agent frameworks.

I also co-founded Alif AI, an agentic EdTech RAG platform, and published VulkanIlm, an open-source Python wrapper for GPU-accelerated local llama.cpp inference on Vulkan-capable hardware. I build with Python, FastAPI, PyTorch, TensorFlow, PostgreSQL, pgvector, Docker, and C++.