Course / Workshop Design • Instructor • Harvard Master Student + CS50 TF/CA
AI Architect • Agentic Engineering • AISP Neural Symbolic Languages • Agentic Sysstem Researcher

I design advanced agentic systems, build symbolic reasoning frameworks, and teach the future of human-AI collaboration. My work blends academic research with hands-on engineering to create adaptive, explainable, and production-ready AI workflows. I study at Harvard Extension School (Digital Media Design, ALM) and instruct across graduate programs and private institutions.

I developed agents that have personality / cognitive traits, deep domain expertise, and self-reflection capabilities. The testing results are strong - and the savant AI team created a research paper. See repo and website below: demonstrate that Savant agents achieve a 56% aggregate quality improvement over standard LLM interactions (87.8 vs 56.2/100)

Repo: https://github.com/bar181/savant-ai-results
Website: Savant AI Testing Results Website
Draft research paper the savants created: Research Paper

I develop modular, adaptive agents powered by symbolic reasoning, AISP, databases, and deterministic templates. These systems reduce ambiguity, accelerate build speed, and support explainable decision loops for production scenarios. Repos include both public tools and advanced private agents for invite-only collaborators.

Currently seeking sponsorship for advanced research. Get early access to leading agentic systems and AI first documentation using a symbolic language designed specifically for AI to AI communication and orchestration.

I specialize in turning complex machine intelligence concepts into practical tools for developers, founders, and teams. My teaching emphasizes smart intern workflows, persona-driven insights, prompt-refine loops, and mixture-of-experts reasoning, giving learners clarity while building real, usable prototypes.

Students create pitch decks, data stories, visualizations, and no-code webapps, using structured reasoning and agentic design to upskill rapidly. Clients include Agentic Learning Labs, Bradley.Academy, and private universities, where I deliver modern AI literacy and hands-on product development experiences.

Here is an example of what we create together: https://github.com/bar181/ai-toolkit

I create robust instructional design systems for technical education, blending Harvard- and Stanford-style teaching models with modern agentic automation. Lessons follow a structured learning arc with clear objectives, consistent voice, and adaptive pacing.

My Agentic Professor agents generate entire courses with tone-tuned lessons:

1. engaging introductions,
2. clear explanations,
3. real-world applications with case studies,
4. activity-driven “show, don’t tell” walkthroughs.

Public repo: https://github.com/bar181/agentic-professor
Private course engine repo: available by request.

(Bradley.Academy — Core Repo)

This system coordinates specialized agents to build end-to-end courses, from learning objectives to assessments. It separates universal knowledge from role-specific applications, allowing content to scale from beginner to master’s level. Adaptive learning ensures consistency with teaching styles I value and model in my own classrooms.

The module architecture automates lesson creation with Bloom-aligned progression and tailored problem sets across experience levels. This is the central workflow I use for workshops, graduate courses, and professional programs.

Private repo — request access

We built this during office hours — an idea that became a full public website using research, algorithms, and API workflows. A real example of turning curiosity into a production artifact.
Public repo: https://github.com/bar181/oscar-win-whisperer

Co-built with students to showcase the exact prompts and workflows used in class for rapid webapp creation.
Public repo: https://github.com/bar181/agentic-lab-ai

A growing collection of persona-driven agents for evaluation, insight, and production workflows. The public repo includes the Dr. House Assessor, built for brutal honesty and rubric-first feedback. Private versions include 30+ specialized agents and 50+ skills supporting Claude Code, the Claude Agent SDK, and modern LLMs.

Public repo: https://github.com/bar181/bar-agents
Private repo: contact for access.

AISP is my research into a modern symbolic language for AI cognition, designed to reduce ambiguity, improve reasoning, and power multi-agent systems with structured memory and deterministic execution. Grammar will be open-sourced. Public repo: coming soon - contact for details. Private repo: contact for details.

A step-by-step onboarding guide for new members learning agentic workflows. Delivered during Agentic Foundation meetups and updated as the ecosystem grows. Public repo: https://github.com/bar181/indiana-meetup

LinkedIn: https://www.linkedin.com/in/bradaross/