CBS SYSTEMS × SIGHTPULSE
Mid to Senior Data & AI Expert
Data Science × Data Engineering
You will make raw e-commerce, CRM, digital and sales & marketing data legible - to humans and to AI agents. This is infrastructure work that ships commercial outcomes.
Belgrade, Serbia (on-site) · Mid or Senior - open · Full-time · E-commerce · Digital · CRM · Pricing
WHY THIS ROLE EXISTS
The layer between raw data and a working agent
SightPulse is a multi-agent analytics and activation platform used by global retailers, consumer brands, and digital-first companies. Specialized agents pull from e-commerce transactions, CRM history, web analytics, pricing feeds, and market data - all at once - to answer commercial questions and activate answers in under 10 minutes.
That only works if the data underneath is clean, well-modelled, and structured so that an LLM can reason over it without hallucinating. That’s you.
Most analytics AI is one layer deep: an LLM on top of a database. SightPulse is different - statistics and ML are the core, AI is the interface. Your job is to build and maintain the gold layer that the agents trust completely.
We are currently scaling SightPulse globally and strengthening the data team for lift-off. This is an opportunity to join at exactly the right moment - when the architecture decisions you make today will shape how the platform runs across dozens of markets.
WHAT YOU WILL ACTUALLY DO
Both sides of the pipeline - or one side deeply
This role sits at the intersection of data science and data engineering, but we are not looking for a perfect 50/50 split. You might come from a stronger modelling background and treat data engineering as a means to an end. Or you might be a seasoned data engineer who has grown into building ML pipelines. Either way, on any given week you could be shaping a segmentation model, refining how a schema gets read by an LLM, or aligning with an agent developer on a data contract. What matters is that neither side is foreign to you.
— Gold Layer Modeling
Design and build the final semantic data layer - clean, pre-aggregated, business-logic-encoded - that feeds SightPulse agents and ML models. Ensure consistency of KPIs and dimensions across all data sources.
— LLM-Ready Data Engineering
Structure tables, schemas, and metadata so LLMs reason over them accurately - no hallucinations from ambiguous column names or missing context. Own naming conventions, documentation, and data contracts with agent developers.
— Metadata Repository
Build and maintain the semantic layer - business definitions, KPI logic, dimension hierarchies. Ensure agents always know what the data means, not just what it contains.
— Predictive & Segmentation Models
Build churn, propensity, CLV, and product affinity models that feed the Customer Engagement Agent. Go beyond RFM - incorporate behavioural, pricing, and digital signals.
— E-Commerce & CRM Data Fusion
Connect GA4, Shopify / platform transactional data, web behaviour, pricing feeds, and CRM records into a unified customer view. Multi-touch attribution logic included.
— Agentic Services Development
Collaborate with agent developers (LangGraph / CrewAI stack) to design data contracts and ensure query performance. Contribute to pipeline design for agentic workflows that run across multiple data sources in parallel.
SENIORITY
Mid or Senior - one role, two speeds
We are open to the right person at either level. Here is how we think about the distinction:
MID LEVEL
Strong executor, growing designer
- 3-5 years in data science or data engineering
- Builds and maintains pipelines independently
- Solid Python and SQL foundation
- Experience with e-commerce, CRM or digital data
- Familiar with ML modelling; growing into LLM tooling
- Executes clearly defined requirements reliably
SENIOR LEVEL
Designs the architecture, owns the outcome
- 6+ years, with clear DS + DE overlap in experience
- Has designed gold layers or semantic layers before
- Understands how LLMs consume and reason over structured data
- E-commerce or CRM domain background - not just generalist
- Defines data contracts; drives architecture decisions
- Mentors and raises the bar for the team
TECHNICAL PROFILE
The stack
Core - must have
- Python (advanced)
- SQL / NoSQL (advanced)
- Propensity / Churn / Pricing Modelling
- Scikit-learn / ML modelling
- Data modelling (star / snowflake schema)
- GA4 / web & digital analytics
- Agentic services development
- Agentic Pipeline Architecture
- ELT / Transformation Pipelines (dbt or equivalent)
Strongly preferred
- Azure Data Lake / Databricks or AWS equivalents
- Medallion architecture (Bronze / Silver / Gold)
- E-commerce data (Shopify, Magento, custom platforms)
- CRM data (Salesforce, SugarCRM, HubSpot, custom CRM)
- CDP / customer data unification
- Attribution modelling
- Pricing data pipelines
- Customer segmentation & CLV
- LangChain / LangGraph
Domain knowledge - strong plus
We work primarily with retail, CPG / consumer, and digital-first businesses across global markets. Experience in one or more of these sectors - understanding how e-commerce funnels work, how CRM lifecycle data is structured, how pricing signals feed commercial decisions - is a genuine differentiator. You do not need to be a domain expert on day one, but you should care about connecting data to real business outcomes.
- Retail / e-commerce / consumer industry experience
- Digital consumer journey analytics
- Trade & sales data
- Metadata / ontology design
- Vector databases (pgvector, Pinecone)
- Data cataloguing (Apache Atlas or equivalent)
WHAT YOU GET
Stability to build on. A stage to disrupt from.
CBS SYSTEMS
20 years of enterprise analytics across 70+ countries and the world’s largest consumer brands. You join a company that has never had to reinvent itself - because it kept investing in what’s next.
SIGHTPULSE
A live AI platform at the edge of what’s technically possible in commercial analytics. Your work ships into production, not a proof of concept. You are here at exactly the right moment.
THE WORK ITSELF
No maintenance queues. No ticketing theater. You design, build, and own the layer that determines whether agents get the right answer. Infrastructure work with a direct line to commercial impact.
WHO WE ARE
Two decades of getting the answers right
CBS Systems was founded in 2007 in Belgrade. Since then we have delivered advanced analytics solutions to leading global brands across retail, CPG, tobacco, FSI, and telco. SightPulse is our flagship AI platform - built on the statistical and ML foundations we have been refining for 18+ years.
18+ Years of advanced analytics | 30+ Data scientists & engineers | 70+ Countries delivered into | 2007 Founded in Belgrade
Trusted by JTI, Ipsos, Coca-Cola, Shell, Delhaize, and 30+ other global brands and regional retailers.
HOW TO APPLY
Send your CV and a short note on what you have built to.
