01 / SYSTEM PROFILEFULL-STACK × APPLIED AI

I build softwarethat can reason,retrieve & ship.

Software Engineer building full-stack and applied AI systems with Python, FastAPI, Next.js, TypeScript, PostgreSQL, and production LLM workflows.

Explore the systems GitHub profile

04Production-minded
case studies

02Surfaces
Web + Android

01Focus
Systems that hold up

BUILDER / 0001

EVIDENCE > HYPE

DISCIPLINEEngineering

CURRENT SIGNAL Building

LLMGROUNDING

DBPOSTGRES

APIFASTAPI

02 / SELECTED SYSTEMS

Built beyond
the happy path.

Four products that move from applied AI investigation to secure SaaS, privacy-first analytics, and cross-platform product engineering.

01 / 04Applied AI / LLM Engineering

TracePilot

Evidence-grounded AI incident investigator

An incident investigation system that turns repository evidence into structured, citation-validated findings—not another ungrounded AI answer.

Hybrid retrieval, durable background execution, validated citations, and production-grade evaluation make every conclusion traceable.

01Incident

02Evidence

03Reason

04Cite

READ-ONLY PUBLIC DEMOThe full investigation pipeline, retrieval, worker, tool calling, and evaluations are implemented and tested.

02 / 04Full-Stack / SaaS Architecture

SupportFlow

Multi-tenant customer support SaaS

A tenant-safe support platform with distinct administrator, agent, and customer workflows built on PostgreSQL row-level security.

RBAC, private attachments, real-time collaboration, ticket operations, SLA tracking, and analytics share one secure data model.

TENANT ISOLATION

ORG_A

ORG_B

ORG_C

RLS POLICY / ACTIVE

03 / 04AI-Enabled Full Stack

SignalRoom

Anonymous feedback & AI analytics

A privacy-first feedback platform that transforms survey responses and private course material into evidence-grounded recommendations.

Single-use response codes, configurable surveys, secure document analysis, dashboards, and AI recommendations work as one product.

PRIVATE INPUTGROUNDED OUTPUT

04 / 04AI Product / Web + Mobile

MuscleBot

AI fitness & nutrition product

A web and Android product combining workout tracking, nutrition logging, progress analytics, and AI-generated planning workflows.

1,300+ exercises, subscriptions, push notifications, Health Connect, and Capacitor packaging take it beyond a browser-only demo.

1.3K+EXERCISES

WORKOUTS

NUTRITION

PROGRESS

03 / ENGINEERING RANGE

From model call
to maintainable system.

I work across the boundaries where AI behavior, backend reliability, secure data, and product experience have to agree.

Applied AI systems

Grounded generation, tool calling, hybrid RAG, structured outputs, reranking, embeddings, and citation validation.

Backend architecture

FastAPI services, PostgreSQL data models, durable queues, multi-tenancy, RLS, retries, leases, and observability.

Product engineering

Polished web interfaces, real-time workflows, analytics, Android packaging, subscriptions, and third-party integrations.

Evaluation & delivery

Adversarial security tests, frozen holdouts, automated testing, Docker, CI/CD, and production-minded deployments.

INPUTCONSTRAINTSEVIDENCEOUTPUT

04 / BUILD PHILOSOPHY

“Useful AI begins where the demo ends.”

The interesting work is not wiring up one model call. It is building the retrieval, validation, security, queues, tests, and interfaces that make the result dependable enough to use.

Evidence-groundedSecurity-awareObservableEvaluated

SIGNAL OPEN

05 / START A CONVERSATION

Have a hard system
worth building well?

I’m interested in applied AI, backend-heavy products, and engineering problems where correctness matters as much as the interface.

LET’S COMPARE NOTES