I build autonomous AI agents, developer tools, and full-stack products — less interested in another chatbot wrapper, more interested in what happens after the first model response: context, memory, tools, delegation, and verification loops that let an agent keep moving without turning autonomy into chaos.
A strong model helps. The system around it decides how much of that intelligence survives contact with a real task.
flowchart LR
A[Objective] --> B[Context]
B --> C[Plan]
C --> D[Execute + Tools]
D --> E[Verify]
E -->|fail| C
E -->|pass| F[Ship]
🤖 Flagship — FRIDAY
A Windows-native autonomous AI agent — voice, screen awareness, desktop/browser automation, persistent memory, multi-agent delegation, and self-verification.
The question isn't "can an LLM answer me?" — it's can a personal AI observe its environment, remember what matters, pick the right tool, coordinate sub-agents, and actually finish the job?
flowchart TD
Core[FRIDAY Core] --> Memory[Memory + Context]
Core --> Router[Tool Router]
Core --> Orchestration[Multi-Agent Orchestration]
Router --> Desktop[Desktop / Browser]
Router --> Dev[GitHub + Dev Tools]
Orchestration --> Workers[Parallel Sub-Agents]
Memory --> Verify[Verification Loop]
Desktop --> Verify
Dev --> Verify
Workers --> Verify
Verify -->|fail| Core
Verify -->|pass| Done[Action Complete]
EmoBoost and Innovation Challenge are early competition builds. They stay pinned-visible because they show the trajectory — not because they're the same caliber as FRIDAY today.
Competition builds → Product experiments → Developer tools → Full-stack systems → Autonomous agents
This profile is curated. Some repos are forks, references, or experiments — the projects above are intentionally selected, not a full inventory.
LLM Agents · Context Engineering · Multi-Agent Systems · Automation · Full-Stack Architecture · APIs · Databases
Phone number stays private — I'll share it directly if a conversation needs it.
The interesting engineering begins after the first model response.