AI Safety Leader & Infrastructure Architect

Bringing enterprise engineering rigor to AI safety.

I’m Bassam. I spent a decade building highly reliable, production-grade cloud infrastructure at companies like Intuit and Zillow. Now, I’m building the open-source infrastructure needed to secure and evaluate advanced AI systems.

You cannot secure what you cannot systematically measure. The AI safety field has strong theoretical threat models, but lacks the battle-tested engineering required to test them at scale. I build the plumbing that makes those evaluations possible.

The Mission & The Bottleneck

Evaluating frontier models for catastrophic risks—such as CBRN capabilities or cyber-offense—remains dangerously inaccessible. Currently, audit-grade tools are locked behind closed doors at frontier labs, while the open-source community relies on uncalibrated, brittle academic scripts.

  • The Problem: AI security cannot rely on manual vibe-checks. It requires the exact same systematic measurement that traditional enterprise software demands.
  • The Goal: To build a dedicated organization focused on auditing frontier risks, starting with open-source, vendor-agnostic infrastructure.
My objective is to democratize AI auditing, empowering independent researchers and policymakers to proactively identify vulnerabilities before deployment.

Track Record: Enterprise Rigor

I know what it takes to keep high-stakes software secure under immense pressure.

Staff Engineer at Intuit

Architected automated testing pipelines that reduced production bugs to fewer than 5 per quarter, fundamentally transforming release velocity.

Staff Engineer at Zillow

Spearheaded the design of end-to-end Continuous Delivery (CD) pipelines serving millions of daily active users with sub-hour MTTR.

Scale & Reliability

Rebuilt legacy event pipelines moving to AWS-native architectures (SQS, Lambda, SNS) to handle massive scale and improve observability.

Dual Capability

Zero-to-One AI Entrepreneurship

I don’t just research problems; I build the products to solve them from the ground up.

CharacBot.com

Founded and architected a full-stack AI SaaS platform, building complex, latency-optimized LLM pipelines from scratch.

Reagent v0.1

Independently built and launched an automated red-teaming framework to standardize how we measure AI vulnerabilities.

The Advantage

Taking zero-to-one ownership as a founder while enforcing the uncompromising security standards of a Staff Engineer.

Current Work & Research

Explore the tools and frameworks I am building to standardize AI safety.

Reagent Framework An open-source CI/CD-integrated platform for automated LLM red-teaming and evaluation. The Risk Spectrum A visual, plain-language guide to understanding AI x-Risk and modern safety frameworks. CharacBot Platform A demonstration of full-stack AI SaaS architecture and optimized LLM integrations. Research Proposals My strategy for building foundational measurement infrastructure for frontier risks.

Let’s secure the future of AI.

I am actively seeking funding, fellowships, and collaborations to scale Reagent and found an independent AI auditing organization. If our missions align, I would love to connect.

Reach out and collaborate