I build language-model agents, computer-vision pipelines, and the backend systems that bring them to users. Based in Lahore, Pakistan.
Currently building production AI.
Studying artificial intelligence at LUMS.
Production systems and open-source work across computer vision, language models, retrieval, and optimisation.
A room photo becomes a furnished design that fits the space. Vision-language models understand the room; an optimisation engine makes the layout physically valid.
I architected the platform, wrote the placement algorithm, and built the image and product-search pipelines.
An open-source harness for small, local language models. It plans, acts, recovers from failures, and checks the result before calling a task complete.
I co-built the execution loop, verification tools, context management, and sandboxed tool integrations.
Helping oncology patients find relevant trials, and researchers explore connections in a drug knowledge graph through plain-language questions.
I contributed to trial matching and built the conversational retrieval system over the research graph.
I’m Saad Hussain Kazmi. I design machine learning systems and the software that delivers them.
I design the model pipeline, build the retrieval and orchestration, and write the backend that brings it to users. My work spans room-layout optimisation, clinical-trial matching, and agents that verify their own output.
Alongside my work as an Associate Data Scientist, I’m pursuing an MS in Artificial Intelligence at LUMS. My work covers model evaluation, retrieval, orchestration, and production APIs.
Professional experience and the foundations behind it.
Retrieval systems, tool-using agents, and workflows with explicit verification.
LangChain / RAG / MCP / n8n
Image analysis, segmentation, multimodal pipelines, and constrained layout search.
PyTorch / OpenCV / vision-language models
APIs, ingestion, vector search, and graph-backed applications.
FastAPI / Django / PostgreSQL / Neo4j
A project, a role, or an interesting problem.
Tell me what you have in mind.