Hi, I'm Marci,
I build AI for regulated industries that an auditor can check: every answer cites its source and every action leaves a trail. I work hands-on inside your team, from the architecture to the conversation with your compliance lead.
Previously led the research team behind Deloitte's reusable KYC credentials. 15+ years building systems where being wrong is expensive.
PhD in computer science,
summa cum laude
A policy, a contract, a fund report: the material your team already works from.
nothing leaves your control
The system answers and attaches the exact passage the answer came from.
click through to the passage
Where the regulator expects a person, approval is a step in the workflow.
a named person signs off
Every action is recorded: who, what, when, and on which evidence.
what your auditor reads
"Is our AI defensible when the regulator asks?"
2-4 weeks, fixed scope
I map one AI system, or your whole estate, against the EU AI Act and ISO/IEC 42001. You get a risk classification, the gaps, and a prioritised roadmap with the evidence to back it.
"Our pilot works but would not pass review."
3-6 weeks
One workflow, built or hardened with the logging, evaluation and provenance that let you prove what it does, then handed to your team with the documentation to run it.
"We need a senior lead, not a full-time hire."
Part-time or full-time, by the month
Fractional or interim CTO or architect: the architecture, the delivery, and the conversations with your auditors and investors.
[PLACEHOLDER: redacted screenshot of the compliance tool, for example a cited answer with its source passage highlighted (16:9)]
Structure-aware chunking, hybrid retrieval with reranking, citation provenance on every answer and an immutable audit trail, orchestrated as Temporal workflows with human approval steps. A Terraform-managed Google Cloud platform of about 180 resources, reviewed plan by plan.
Python · Docling · LlamaIndex · pgvector · Temporal · Terraform · Google Cloud
[PLACEHOLDER: one sentence from Nick's Upwork review, job "Docling RAG Pipeline: Smart Chunking & Reranking"]
Every answer traces to a passage in the source document.
Human approval is a step inside the workflow, not a policy.
Case study in preparation, published with the client's sign-off.
[PLACEHOLDER: image of you on the KILT work, for example a still from the Built with Substrate interview (16:9)]
I led the research team that took Deloitte's reusable KYC credentials to production on the KILT network: system architecture for the self-sovereign identity protocol through its mainnet launch on Polkadot, a compliant token sale, and the KYC and KYB credential that Deloitte issues on it.
Polkadot · decentralized identity · token economics · Kubernetes
[PLACEHOLDER: one sentence from Christine Mohan's LinkedIn recommendation]
Live on mainnet since 2021.
Deloitte issues the credentials.
Sources: Finextra, Deloitte on LinkedIn, KILT Protocol, and the Polkadot Decoded 2023 talk.
[PLACEHOLDER: redacted screenshot of the coaching app or its rules file (16:9)]
A four-agent coaching backend with dynamic prompt assembly, journey-gated tools, a safety mode for overwhelmed users and full tracing through Langfuse. The coach's rules moved out of code into a file the team owns.
Python · multi-agent LLM backend · Langfuse · OpenTelemetry
[PLACEHOLDER: one sentence from the TriggerCoach founder's Upwork review, job "Chatbot MVP Development & System Architecture"]
Two weeks to days per new feature.
Zero deployments to change what the AI says.
Full write-up being moved to this site.
[PLACEHOLDER: redacted screenshot of a WhatsApp memo or the Airtable view (16:9)]
Airtable, n8n and the WhatsApp Business API, with an LLM memo writer wrapped in a code-level guard: every figure is rendered from a field, never generated. Readiness-gated announcements and commitment tracking, live in production.
Airtable · n8n · WhatsApp Business API · LLM with a code-level guard
[PLACEHOLDER: one sentence from the fund (to be collected), with how they want to be named]
No number in a memo comes from the model.
Live with real investors since August 2026.
Reference available on request.
I'm Márton Csernai, Marci to most people (say MAR-chee). I hold a PhD in computer science, summa cum laude, and have spent fifteen years building systems where being wrong is expensive: data-centre networks, neuroscience data pipelines, a self-sovereign identity network, and now AI for regulated industries.
I work hands-on, inside your team. I write the architecture, run the infrastructure, ship the code, and sit in the room with your compliance lead. What I don't do is policy slideware. If an AI system cannot produce the evidence an auditor asks for, I don't consider it done.
Python · TypeScript · RAG · agents · evaluation · Temporal · Terraform · Google Cloud · Kubernetes
Berlin and London, remote across the EU and UK. Contract, interim or fractional. CV on request. Hire me on Upwork, where I'm Top Rated with 100% job success.
[PLACEHOLDER: one sentence from the Upwork review of the "AI Workflow Automation for Internal Operations" job]