My advantage is deep business and process understanding: knowing how work gets done, where it gets stuck, and what is worth building. I engineer business systems by choosing the right combination of deterministic software, AI, and human judgment.
AI gives me execution leverage. Business understanding sets the direction.
- Observe the real work. Map the people, tools, handoffs, exceptions, and constraints before proposing a system.
- Route each step. Use software for rules and calculations, AI for bounded interpretation and drafting, and people for accountable decisions.
- Design for failure. Make duplicate work, missing data, bad output, outages, and recovery visible. Define fallbacks and escalation owners.
- Verify against real historical examples. Reconcile with source records and operator judgment before rollout. Synthetic fixtures prove engineering behavior; they do not establish business acceptance.
- Measure business value. Establish a baseline, track the change, and include operating costs. Separate recovered capacity, realized savings, and financial exposure.
This is my delivery method. The projects below demonstrate its engineering choices; their case studies distinguish implemented behavior from business validation still to be done.
| Project | The business-system judgment it demonstrates | Evidence and scope |
|---|---|---|
| Conduit-OS | Reliable commerce event handling before downstream automation; deterministic fraud rules, replay, role boundaries, and read-only financial queries | FDE case study · local prototype and historical simulator results |
| Executive Daily Brief | Software calculates findings; optional AI narrates; executives decide what to investigate | FDE case study · fictional data, deterministic fallback, assumptions-based value model |
| Public Companies Data Scraper | Deterministic SEC/XBRL normalization, bounded AI screening, and analyst due diligence | FDE case study · saved public filing fixtures; extracted figures remain unverified |
| Customer Support Agent | AI classifies and drafts; a person checks policy, handles exceptions, and sends the response | FDE case study · draft-only n8n prototype; integration acceptance remains open |
| The Conductor | Deterministic caching, provider failover, usage accounting, and explicit latency/correctness tradeoffs | FDE case study · historical deployed benchmarks with mock providers and documented limits |
| CleanWrite | Local document ownership, deterministic analysis, and optional AI suggestions that require review before application | FDE case study · local product implementation with replacement-safety tests |
These are inspectable engineering projects, not claims of client deployment, completed shadowing, production adoption, or measured commercial ROI.
- Paid Systems Opportunity Audit: understand the current workflow, establish the baseline, and produce a prioritized systems roadmap.
- Build & Implementation: scope, engineer, verify, and introduce the selected system.
- Systems Management: monitor, maintain, and improve the system against agreed operational and business measures.
A free fit call is an initial conversation; it is separate from the paid audit. KylixAI · Website source
TypeScript · Python · SQL · Next.js/React · FastAPI/Fastify · Postgres/Supabase · SQLite · Redis/BullMQ · n8n · Trigger.dev · API integrations · structured model outputs · automated tests
Open to Forward-Deployed Engineering roles and scoped business systems engagements.