Hi — I’m Mohammad Zoraiz, a student at Duke University studying Electrical & Computer Engineering, Physics, and Computer Science. I work across agentic AI systems, biomedical intelligence, quantum computing, embedded hardware, and trustworthy infrastructure.

I am currently working with Amazon AGI Lab and Amazon Kiro Labs, where I focus on low-latency voice-to-agent workflows and real-time coding-agent infrastructure. Previously, I worked on AI for genomics with the Pfizer Computational Genomics team, where I helped advance modeling of the human genome for drug discovery, and on agentic AI and LLM evaluation infrastructure at Amazon Web Services, including benchmarking platforms, serverless analysis systems, and tooling used across Amazon Q, Kiro, and AWS Agentic AI teams.

Before that, I worked in embedded systems on ControlLogix at Rockwell Automation, researched quantum algorithms through the MIT Center for Quantum Engineering with MIT Lincoln Laboratory, worked with LifeEdit on a breast cancer cell-classification methodology, and built systems spanning spatial proteomics, precision oncology, gene-editing classification, psychoacoustic detection wearables, and privacy-preserving computation. Across these experiences, I have become most interested in building AI systems that are not only capable, but also auditable, reliable, and useful in domains where mistakes matter.

I am an AI systems researcher and builder. I am interested in how intelligent systems can reason over messy, high-stakes information: scientific evidence, clinical data, software traces, physical sensors, and human context. My work follows three connected themes: 1. making AI systems more agentic and useful; 2. grounding predictions in evidence, provenance, and uncertainty; and 3. connecting computation to the physical and biomedical world.

• Biomedical and Scientific AI: I build AI systems for precision medicine, drug discovery, and genomic analysis. My work includes Provenia Bio, a proof- and provenance-carrying neuro-symbolic drug discovery system designed to make therapeutic hypotheses auditable and machine-checkable; TargetONCO, an agentic precision oncology platform combining radiology, spatial proteomics, clinical retrieval, and Bayesian differential ranking; and gene-expression classification work for edited versus unedited biological signatures. More broadly, I am interested in AI systems that can help scientists move from raw data to traceable, experimentally meaningful hypotheses.
• Agentic AI and Infrastructure: I build infrastructure for agents that code, reason, remember, and improve over time. At AWS and Kiro, I have worked on benchmarking and runtime systems for iterative LLM improvement and coding-agent execution. My independent work includes ctrlSlash, an open-source documentation and context engine for AI IDEs, and ombench, a memory and backtesting framework for operational agents that evaluates whether compiled context actually improves performance on historical tasks. I care about agent systems that are not just impressive in demos, but measurable, reproducible, and dependable.
• Quantum, Trust, and Secure Computation: I work on computational systems where structure matters. My quantum work includes a Quantum Bayesian Learner with Hardware-Aware Circuit Compression, studying how quantum models trade off accuracy against realistic hardware cost. I also built CipherShield, a privacy-preserving aggregation system using homomorphic encryption and blockchain verification, with the goal of making sensitive-data misuse mathematically impossible rather than merely policy-prohibited.
• Embodied and Human-Centered Hardware: I like building systems that touch the physical world. SonicSync is a submersible wearable apparatus developed with Garmin engineers for underwater psychoacoustic feedback and testing. Qadam is an accessible biotech venture that imagines a portable prosthetics clinic in a case, combining open hardware, offline fitting workflows, and field-ready clinical tools. I have also built autonomous and agentic sensing systems, including pollination-drone simulation and computer-vision infrastructure.

Apart from research, I build products and ventures around a simple principle: powerful systems should become more accessible, transparent, and useful.

Through Qadam Labs and independent projects, I work on AI-native tools for developers, researchers, clinicians, and organizations. Some of these projects are infrastructure-heavy, such as agent benchmarking, documentation retrieval, vector search, and cloud-native ML pipelines. Others are more human-facing, such as education technology, accessible prosthetics, biomedical decision-support systems, and tools for communities that are often underserved by advanced technology.

The common thread is that I like building full-stack systems end-to-end: from the model and algorithm, to the backend architecture, to the interface, to the deployment path, to the question of whether the system actually helps someone make a better decision.

Beyond research and building, I care a lot about living with curiosity, discipline, and a sense of play. I train for marathons and triathlons, and I am drawn to endurance sports because they reward patience, consistency, and the ability to keep moving when things become difficult. I also enjoy alpine climbing, especially the mix of physical challenge, technical judgment, and quiet focus that comes with being in the mountains.

I love conversation in all its forms: long walks, philosophy debates, late-night arguments about ideas, and the kind of talking that helps people understand each other more deeply. I am also interested in learning new languages, both as a way to connect with people and as a way to see how different cultures organize thought, emotion, and meaning.

Creatively, I enjoy photography, writing poetry, design, and board games. Photography helps me notice the world more carefully; poetry gives me a way to make sense of what I feel; and board games scratch the same itch as systems design, strategy, and debate. Across all of these, I am usually looking for the same thing I look for in my work: beautiful systems, meaningful questions, and better ways to understand people.