Services

What I do well, backed by shipped projects.

AI-Native Engineering

What: Multi-agent systems, LLM orchestration, custom Claude pipelines. The engineering itself runs through AI — design, code, tests, documentation, review — not just the product surface.
For whom: Founders and teams that want AI-first workflow from day one, not an AI feature bolted onto a traditional codebase.
Examples: FirmPilot — nationwide pipeline transforming US court opinions across all states into structured, LLM-ready data. Majesti AI — a Telegram product where every feature, including a three-tier referral system, was vibe-coded end-to-end through Claude. An open Claude Code orchestration setup that colleagues use as a reference.
Engagement: Four to twelve weeks, fractional or project-based.

What: Lead a small R&D team from a working proof to a production-ready first release. Architecture, hiring decisions, test discipline, handoff documentation.
For whom: Startups with engineering capacity that need a senior technical lead — fractional CTO style — without committing to a full-time hire.
Examples: OneLiner — Heroku-style PaaS on Kubernetes built with a small team on a research grant. FirmPilot — pipeline architecture and test discipline that let other engineers add new state parsers in hours instead of days. Toptal conversion engine — six years tech-leading the engine, then handing it off to a ten-person team that owned it for years afterward.
Engagement: Two to three days per week, three to nine months.

What: Translate between business stakeholders and engineers. Production-readiness reviews, technical due diligence, architecture audits delivered in language each side actually uses.
For whom: Companies where business and engineering have stopped speaking the same language — pre-investment audits, post-launch stabilization, cross-functional rescoping.
Examples: ASIA Consulting Group / OneBox — three years embedded in a management-consulting team as solo R&D lead, running company diagnostics, strategy sessions, and HR assessments alongside building the technical side (later acquired by Ernst & Young). Presidential Management Program (RF, type Q, 2010) — formal training in management theory and cross-disciplinary translation, complementing the engineering track.
Engagement: Workshops plus ongoing advisory; typical engagement is four to eight weeks.

Every engagement starts with a short conversation. Tell me what you're building and I'll tell you whether I can help — and how.

Toptal handles contracts and payment. Or book a 30-min intro call