Research and Development

Our R&D work at Teknopark.

97 TEAM is a technology company located at Gaziantep Teknopark. We conduct our research and development activities through defined research questions, measurable success criteria and written outputs, and we transfer mature results directly into our products and client projects.

Our Approach

We do R&D inside real projects.

The purpose of our research work is not to produce academic publications but to develop reusable solutions to the real engineering problems we encounter in client projects.

Every research topic begins with a work plan setting out the problem definition, a review of the current state of the art, the hypothesis, the experimental design and the evaluation criteria. Studies are reviewed periodically, and unsuccessful hypotheses are also recorded together with the reasons. This discipline ensures that the knowledge gained belongs to the organisation rather than to individuals.

Mature outputs become internal libraries, reference architectures and development tools, so that each new client project starts faster and more safely than the one before it.

R&D PROFILE

Location Gaziantep Teknopark, Technology Development Zone
Research staff Specialists from the software architecture, data science and quality engineering disciplines
Focus areas Applied data analytics · Document understanding · Distributed systems · Quality automation
Working standards Version-controlled experiment records, code review and reproducibility as a founding principle
Output formats Technical reports, reference architectures, internal libraries and prototypes

Research prototypes are written with the same discipline as production code: version control, code review and automated tests.

Research Focus

Four principal research areas

We select our focus areas according to the problems we encounter most often in client projects and for which existing off-the-shelf solutions fall short.

Primary research area

Applied Data Science

The broadest heading in our research programme. We work on data-driven components that can genuinely be integrated into enterprise workflows, with predictable cost and auditable behaviour.

Consolidating scattered enterprise data into a single source of truth
Verification of model outputs and measurement of error margins
Optimisation of query cost and latency
Evaluation pipelines and regression testing

Document Understanding and Information Extraction

An organisation's largest source of data is its unstructured documents. We focus on extracting information from those documents reliably, traceably and auditably.

Parsing complex page layouts and tables
Text and field recognition in scanned documents
Linking extracted fields back to the source document
Domain-specific language processing for Turkish texts

Distributed System Architectures

We work on architectural patterns that meet high availability and low latency objectives while keeping operating costs under control.

Consistency patterns in event-driven architectures
Data isolation in multi-tenant systems
Automatic scaling and the cost-performance balance
Fault isolation and graceful degradation strategies

Software Quality Automation

We develop methods that reduce manual effort in quality assurance while increasing coverage. The aim is a safety net that catches defects before release.

Automation in test case generation
Identifying defect-prone code through static analysis
Automation of visual regression and accessibility checks
Quality gates embedded in the release pipeline

Research Process

How an idea becomes a product

Every research topic must meet defined criteria in order to progress to the next stage. Work that does not meet the criteria is archived together with the reasons.

GATE 01
Problem Definition

A problem arising from client projects or from observation of the sector is set down in writing; existing solutions and their limitations are reviewed. Output: a problem statement.
GATE 02
Proof of Concept

The core hypothesis is tested with a limited-scope prototype. Success criteria are set in advance. Output: an experiment report and a decision note.
GATE 03
Engineering Maturation

The prototype is brought to production standard in terms of error handling, observability, security and performance. Output: a reusable component and technical documentation.
GATE 04
Field Validation

The component is used under controlled conditions in a real client project; measurements and feedback are collected. Output: a field evaluation report.
GATE 05
Standardisation

The validated solution is incorporated into our corporate reference architectures and development standards, and the whole team is trained. Output: a standard and training material.

Infrastructure and Capability

The infrastructure we use

We operate our development, experimentation and validation environments separately, under version control and in a reproducible manner.

Model Development

Python · pandas
PostgreSQL · data warehouse
scikit-learn
Experiment tracking and versioning
Reporting and visualisation

Data Engineering

Python · Pandas · Polars
Transformation pipelines (ETL/ELT)
PostgreSQL · Elasticsearch
Data quality checks
Anonymisation methods

Cloud and Operations

Microsoft Azure
Amazon Web Services
Docker · Kubernetes
Infrastructure as code (IaC)
Monitoring and cost management

Quality and Security

Automated test pipelines
Static code analysis
Dependency and licence auditing
Secret scanning
Access control and audit logging

Collaboration

Open to joint work

The greatest benefit of being located in a technology park is the ability to conduct research within an ecosystem rather than in isolation.

Discuss a collaboration

University collaborations

We are open to internships, dissertation work and joint projects with undergraduate and postgraduate students. Students work on real problems with mentoring support.

Industry partnerships

We are open to joint R&D proposals from organisations in manufacturing, logistics, healthcare and retail, based on real field data; scope and data sharing are defined by contract.

Technology provider programmes

We follow providers' technical programmes for cloud and data infrastructure, and actively use these platforms in client projects.

Open source contribution

We are committed to releasing tools of general benefit that are not client-specific as open source, and to contributing to the projects we use.

Intellectual property management

In joint work, the sharing of intellectual property, publication rights and confidentiality terms are defined in writing at the start of the project.

Are you considering a joint research project?

Whether you are a university, an industrial organisation or a technology provider, we would be glad to consider your proposal for collaboration. Let us begin with a short introductory call.

Get in touch zai@97.team