Computer Science student at CUNY building production software, multi-tenant SaaS, AI systems, and full-stack apps.

Most of my work lives at the intersection of backend engineering, applied AI, and security. I care about the parts that don't demo well: tenant isolation, migration safety, request-boundary validation, and accessibility.

Multi-tenant SaaS for Amazon DSP delivery operations. Replaces spreadsheets and verbal coordination with a role-aware platform covering the full shift lifecycle.

• Scale: 688 commits · 582 backend tests · 41 API routers · 136 migrations · 8 roles
• Stack: FastAPI · PostgreSQL · SQLAlchemy 2.0 · Redis · Celery · React 19 · React Native · Docker · AWS (EC2, Cognito, Textract)
• Highlights: weighted dispatch algorithm (preferences, PTO, trainer–trainee pairing, crew balance) · two-tier package routing · Discord bot integration · SSE live updates · ADP payroll reconciliation · strict company_id tenant isolation on every query
• Status: deployed to staging on AWS with CI/CD via GitHub Actions

Fans a query out to multiple LLM and search providers in parallel, then ranks results with semantic embeddings and three-axis scoring (price · distance · rating) — and demotes sponsored content instead of rewarding it.

• Stack: Python · NLIP protocol · OpenAI · Gemini · Ollama · WatsonX · Brave Search · Google Places · Prometheus
• Highlights: parallel fan-out with failure isolation · constraint extraction from natural language · fuzzy consensus clustering · sponsored-content penalty · 277 passing tests

Full-stack app that analyzes skin conditions and generates personalized, budget-aware skincare routines.

• Stack: React · FastAPI · Gemini 2.0 Flash · privacy-focused image pipeline

Agentic AI academic advisor built for the AI Innovation Challenge. Checks visa status, financial aid eligibility, and graduation timeline before recommending a course — the constraints DegreeWorks doesn't model.

Tracked how Google Play review sentiment shifted in response to real-world events affecting TikTok. Python · Pandas · Matplotlib.

Languages: Python · TypeScript · JavaScript · SQL · Kotlin

Backend: FastAPI · SQLAlchemy 2.0 · Pydantic · PostgreSQL · Redis · Celery · Alembic · REST · SSE

Frontend & Mobile: React 19 · React Native · Tailwind CSS · Vite · Jetpack Compose

Cloud & DevOps: AWS (EC2, Cognito, Lambda, Textract, SSM) · Docker · GitHub Actions · Caddy

AI / ML: LLM orchestration · semantic embeddings · RAG patterns · prompt engineering · OCR pipelines

Security: multi-tenant isolation · JWT/JWKS auth · role-based access control · OWASP Top 10 · PII retention & privacy review

Security on my projects isn't a separate hobby — it's the audit I run before calling anything done. On AsheFlow that means a documented nine-dimension review of every change: tenant scoping on every query, role gates on every endpoint, no unvalidated Any at the request boundary, no exception text leaking to clients, and no PII in logs or response schemas. I'm extending that instinct toward penetration testing and offensive security.

• Shipping AsheFlow toward production launch
• Building Angel Filter for a May 2026 capstone demo
• Going deeper on distributed systems, system design, and offensive security

Open to software engineering and security internships/roles, and to collaborating on ambitious projects. If you're working on something hard, I'd like to hear about it.

_{Thanks for visiting 👍🏾 — let's build something.}