About Sasa Fajkovic.

I’m an Applied AI Lead and Staff Engineer based in Berlin. I set engineering direction, grow teams and connect technical decisions to business outcomes, while staying hands-on with architecture and production systems.

Download CV (PDF) Connect on LinkedIn

The work after the AI demo.

I work at both organisational and technical depth, from direction and team growth to architecture, delivery and production AI.

Applied AI

Design and build production systems with defined responsibilities, scoped access and work that can be evaluated.

Engineering leadership

Set technical direction and grow teams while keeping ownership and accountability clear.

Hands-on building

Stay close to architecture, code and operations, from open tools to self-hosted infrastructure.

I also founded Changify, a Berlin consultancy helping organisations improve processes and adopt AI with software they can own long-term.

Topics I can shape for your audience.

These are practical talk directions that can be adapted to a conference session, podcast or leadership roundtable.

Managing hybrid teams of people and AI employees

For: Leaders introducing AI agents into day-to-day operations

Moving from an impressive agent demo to a defined role that human colleagues can understand, evaluate and trust.

Attendees will leave with

Define an AI employee's responsibilities and access before choosing its tools
Evaluate the work with deterministic checks and human review
Keep human ownership visible as people and AI employees work together

Giving agents production access

For: Platform, security and engineering leaders

Giving an agent enough access to be useful without losing control of approvals, evidence or recovery.

Attendees will leave with

Scope permissions around a job rather than a broad capability
Place approval boundaries where consequences become material
Design audit and recovery paths before increasing autonomy

AI infrastructure you can operate and own

For: AI, platform and technical decision-makers

Choosing infrastructure that keeps cost, control and operational responsibility understandable after launch.

Attendees will leave with

Map workloads to infrastructure instead of following a default stack
Compare managed and self-hosted options through cost and control
Treat monitoring, failure and maintenance as design inputs

Autonomous feedback and self-improvement for AI systems

For: AI, platform and product teams operating AI in production

Building feedback processes that help AI systems improve from real outcomes without obscuring ownership or quality decisions.

Attendees will leave with

Separate feedback signals from decisions about system changes
Evaluate proposed improvements before promoting them
Keep auditability and human accountability visible throughout

Scaling engineering teams with clear ownership

For: Engineering managers, directors and growing product teams

Growing delivery capacity without making responsibilities, decisions and technical direction harder to understand.

Attendees will leave with

Make decision ownership explicit as the organisation grows
Scale the operating model alongside team size
Use structure to support autonomy rather than adding ceremony

Leadership through technical and organisational change

For: Engineering leaders and cross-functional teams

Leading change while keeping the purpose, trade-offs and expectations clear for the people doing the work.

Attendees will leave with

Connect technical choices to outcomes people can understand
Create feedback loops before a change becomes difficult to reverse
Keep accountability visible across team boundaries

Human drive and a culture of proactive feedback

For: Engineering leaders, managers and teams navigating change

Creating a culture where people actively ask for, give and act on useful feedback instead of waiting for formal review cycles.

Attendees will leave with

Make feedback routine, timely and specific
Model the proactive behaviour expected from the team
Connect individual growth with stronger team outcomes

Business process optimisation before automation

For: Business, operations, product and technology leaders

Improving a workflow before software or AI makes its existing friction faster and harder to change.

Attendees will leave with

Map how work happens before choosing a tool
Remove unnecessary steps and hand-offs before automating
Measure the outcome that the redesigned process should improve

For event organisers.

My first-person background, current headshot and LinkedIn profile for event planning and introductions.

Download JPG · 1181 × 1074 · 197 KiB Download WebP · 1181 × 1074 · 83 KiB

Short biography, in my own words

I lead Applied AI at Trade Republic, where I am also a Staff Engineer. Based in Berlin, I set technical direction, grow engineering teams and stay hands-on with architecture and production AI systems. I share practical lessons about leadership, hybrid human and AI teams, infrastructure ownership and responsible adoption. I’m also the founder of Changify and the creator of agents-skill-eval and unraid-rsync.

Long biography, in my own words

I’m an Applied AI Lead and Staff Engineer based in Berlin. At Trade Republic, I set technical direction for applied AI, grow the team and work hands-on with the architecture and software behind production systems. My work connects engineering, product and business outcomes. I focus on the leadership, evaluation practices and governance needed for people and AI employees to work together responsibly. Before focusing on applied AI, I led engineering teams in banking, built B2B and B2C software products and taught programming and productivity courses to ~500 people. I founded Changify, created the open-source projects agents-skill-eval and unraid-rsync, and spoke at WeAreDevelopers World Congress 2026.

Base: Berlin, Germany.

Connect on LinkedIn

Professional experience.

Seven roles spanning applied AI, financial technology, product development, teaching and business operations.

2026 — Present

Founder

Changify · Berlin, DE

AI, process & software consultancy helping organisations ship production software they can own long-term.

Process optimisation — mapping real workflows and redesigning them for simplicity
AI transformation — goal-mapped adoption, custom tooling and cost control
Software delivery — backend systems (JVM, Go), APIs and full-stack applications
Advisory — technical due diligence and build-vs-buy strategy

2022 -> Present

Applied AI Lead, Staff Engineer

Trade Republic · Berlin, DE

Defining how the organisation adopts and operationalises AI across engineering, product and business.

Building and leading the Applied AI team — hiring, growing and setting technical direction
Shaping the AI operating model, governance, evaluation frameworks and responsible use practices
Driving LLM and AI integration across the product surface
Cross-functional leadership across product, engineering, operations and business

2020 — 2022

Engineering Manager

Deutsche Kreditbank AG · Berlin, DE

Led a cross-functional team across backend, frontend and QA.

Delivered new customer onboarding flows and provisioned 5+ million customers to a new debit card
Replaced legacy systems with modern HTTP REST APIs
Kotlin, Spring Boot, Kubernetes, Docker, Prometheus, Grafana, Kibana, GitLab CI and PostgreSQL

2018 — 2020

Founder

Digitalni val · Zagreb, HR

Built and operated B2B and B2C digital products.

A proprietary real-time event-driven betting system
E-commerce sites, websites and proprietary B2C CRM tooling
Retail office management
Kotlin, Spring Boot, Node.js, Puppeteer, Ubuntu, Windows Server and PostgreSQL

Apr 2016 — Jan 2020

Lecturer

Algebra · Zagreb, Croatia

Taught ~500 people across programming and productivity courses.

Web development with PHP, HTML, CSS and MySQL
Java programming for backend and Android
Microsoft Excel and Access from beginner to ECDL Expert

Jan 2016 — Jun 2019

Engineering Manager

UniCredit Group · Zagreb, Croatia

Led a full-stack engineering team delivering core retail banking products.

Built CRM tooling and Do Not Call registry integration
Delivered the PSD2 backend implementation on time
Java, Spring Boot, HTML, CSS, JavaScript, AngularJS, Jenkins, GitLab, IBM CICS and DB2

2014 — 2016