Useful beats impressive
AI should solve a real workflow problem, not only demonstrate technical possibility.
Production AI, without theatre.
Tamas Darvas
AI Engineering Lead for Production AI Platforms in Zürich
I help enterprise teams move AI from prototype to production - where secure data, real evaluation, clear ownership, and engineers who stay with the system matter.
- 30+
- years in IT, including 17+ in financial technology
- Platforms
- production AI, data, and software systems
- Leadership
- hands-on technical direction, mentoring, and delivery
Production AI lens
Production AI needs more than a good model.
A convincing AI demo is useful, but it is not the system. The harder work starts when a team has to answer practical questions: who can see which documents, how quality is measured, who owns failures, what gets logged, and how the next team ships safely on the same platform.
Governance belongs in the architecture
Access control, auditability, and policy should be designed into the system, not added after launch.
Platforms scale better than heroics
Reusable patterns, paved roads, and developer experience help teams ship safely without depending on a few specialists.
People make AI production-ready
Adoption, ownership, mentoring, and stakeholder alignment decide whether the system survives beyond the prototype.
Production AI framework
AI, people, data, and code - designed together.
The useful work is not choosing one dimension. Model behavior, human ownership, trustworthy context, and reliable delivery have to reinforce each other from the start.
Model behavior
LLMs, RAG, agentic workflows, evaluation, and model integration.
Ownership and adoption
Leadership, mentoring, stakeholder alignment, adoption, and ownership.
Trustworthy context
Retrieval, access control, data quality, lineage, freshness, and governance.
Reliable delivery
APIs, platforms, observability, reliability, developer experience, and integration.
What I help solve
AI engineering problems I help teams solve.
From AI prototype to production platform
Turning promising LLM, RAG, and agentic AI experiments into secure, observable, maintainable systems.
Enterprise RAG and knowledge systems
Building document-grounded AI that respects permissions, shows where answers came from, measures weak answers, and gives the team a feedback loop they actually own.
Agentic workflows with guardrails
Putting agents behind explicit tool permissions, human approval steps, audit trails, and fallback paths before giving them more autonomy.
AI platform developer experience
Turning common AI work - model access, retrieval, evaluation, observability, and rollout - into paths teams can reuse.
Data and model-serving foundations
Making the handoff from data pipelines to APIs and model-serving boring enough that product teams can rely on it.
Leadership when the AI work is unclear
Helping engineers and stakeholders decide what should be an experiment, product, or platform before the team commits to delivery.
Differentiation
What I bring to production AI teams.
Long-range judgment
30+ years of IT work give me a practical sense of what survives production and what becomes operational debt.
Regulated context
17+ years in financial technology mean security, reliability, and governance are design constraints from the start.
Hands-on leadership
Architecture reviews, code-level trade-offs, mentoring, and delivery planning stay connected instead of drifting into separate conversations.
Platform thinking
The goal is not one impressive AI feature. It is a delivery path other teams can reuse safely.
Clear communication
Compliance, product, engineering, and business stakeholders get the same trade-off in language they can act on.
Recruiters and hiring managers
For recruiters and hiring managers in Zurich.
If you are hiring for a Zurich AI engineering lead who can review architecture in the morning, unblock engineers in the afternoon, and keep regulated delivery on track, let's talk.
Profile
AI Engineer in Zürich with people management and hands-on technical depth.
I am an AI Engineer and Engineering Lead in the Zürich area with 30+ years in IT and 17+ years in financial technology. My work combines hands-on development, technical direction, and people leadership in teams that have to ship securely and reliably.
Current work includes LLM integration, RAG, agentic workflows, data pipelines, model-serving infrastructure, and developer-facing tooling built with the engineers who operate them.
I work where AI ideas become owned software: the data is permissioned, quality is measured, engineers can operate it, and the next team has a path to ship.
Professional Experience
People leadership and hands-on engineering across AI platforms, data systems, and business-critical applications.
Feb 2024 - Present
Director - AI Engineering Lead, AI Platform
UBS, Zurich
- Lead, mentor, and grow a chapter of AI and software engineers advancing the organization's AI platform capabilities.
- Stay hands-on in architecture and implementation, from platform design through production-grade AI tooling and working code.
- Drive program planning, dependency management, stakeholder alignment, and delivery planning across product streams.
- Translate technical trade-offs into business-level decisions for senior leadership.
- Define AI engineering standards for LLM-powered applications, Agentic AI, RAG architectures, agentic systems, and responsible AI integration.
Feb 2022 - Jan 2024
Vice President - DWH Data & Software Engineering
Credit Suisse, Zurich
- Managed end-to-end delivery of data platform modernization work in a regulated financial-technology environment.
- Coordinated workstreams across engineering teams and business analysts.
- Ran cross-functional dependency routines, risk registers, prioritization, and stakeholder communications.
- Led data integration work with React, Java, Oracle, Linux, CI/CD, DevOps, and scalable platform migration.
Jul 2007 - Jan 2022
Vice President - Software Engineering, App Owner
Credit Suisse, Zurich
- Owned the full software lifecycle for business-critical trading and regulatory applications.
- Managed onshore and offshore teams across Zurich, Budapest, and Pune.
- Led long-running application modernization spanning front end, back end, data, and infrastructure releases.
- Established CI/CD and DevOps practices that improved deployment lead time and production stability.
1996 - 2007
Earlier Engineering Professional and Leadership
Alcoa, TATA Consulting, PromonIT, IBM
Senior Oracle Architect, Solution Architect, SAP team leader, and department management roles that built the foundation for later architecture and delivery leadership.
Capabilities
Where people leadership, architecture, and hands-on engineering meet.
People & Engineering Leadership
- People management
- Mentoring and coaching engineers
- Engineering chapter leadership
- Stakeholder management and influence
- Risk identification and mitigation
- Multi-team coordination
- Agile delivery and process optimization
AI & Data Engineering
- Hands-on development
- LLM integration
- RAG architectures
- Generative AI
- Agentic workflows
- ML pipelines
- Data warehousing
Software Platform Stack
- Java
- Python
- JavaScript / TypeScript
- SQL / PL/SQL
- React / Next.js
- Azure Cloud
- CI/CD and DevOps
- Bash and *nix
Credentials
Continuous AI learning on top of a systems engineering base.
Education
- MSc, Applied Data Science and AI
- BSc, Banking and Finance
- BSc, IT Engineering
Recent AI Certifications
- Google Cloud Gen AI Agents: Transform Your Organization
- Google AI Professional
- IBM Generative AI Engineering Professional Certificate
- DeepLearning.AI Generative AI for Software Development
- Oracle Cloud Infrastructure Generative AI Professional
Engineering Foundation
- UBS Certified Engineer Gold
- Microsoft Certified Azure Data Engineer Associate
- Oracle Database Administrator Certified Professional
- Oracle Certified Java Programmer
- Frontend Developer, freeCodeCamp
Building AI beyond the prototype stage?
If the work involves RAG, LLM integration, platform foundations, governance, or senior hands-on engineering leadership in Zurich, send me the context.
Contact
Based in Adliswil, working across AI engineering, people leadership, and hands-on software development.
Swiss national. Fluent in English, native Hungarian, German B1. Outside work, I am usually on a hiking trail, running route, motorcycle, or building something for the web.