SOFTWARE ENGINEER · ENTERPRISE SYSTEMS · PAYMENTS · DOCUMENT SYSTEMS · AI

I build the systems a bank cannot afford to break.

Complexity is inevitable.
Bad architecture isn't.

ISO 20022 payments, document management, applied artificial intelligence: I build the part of the software that has to stay reliable when everything else fails. Currently Software Engineer at the Bank of Algeria.

Explore my work Get in touch

THE LIVING ARCHITECTURE · SCROLL TO EXPLORE

02 / THE ENGINEER BEHIND THE SYSTEM

An engineer who designs, integrates and evolves complex enterprise systems.

I'm Salim Mabed, a Software Engineer trained in Intelligent Computer Systems, with experience in environments where reliability, security and precision are non-negotiable.

My path has taken me from software development and artificial intelligence through to enterprise systems: payment systems, financial and accounting integration, digitisation and electronic archiving.

What interests me most are architectures, flows, integrations and the reliability, security and traceability requirements that let several systems work as one. But a high-performing system does not have to be hard to use, which is why I give technical architecture and user experience equal weight.

My goal is simple: build technology that works beautifully, inside and out.

3
YEARS · DEV & AI
5
PROJECTS SHIPPED
12
CERTIFICATIONS

03 / EXPERTISE

Six domains, one thread: keeping enterprise complexity under control.

Enterprise Systems Engineering

01

Designing and integrating enterprise systems that stay reliable, secure and auditable in regulated environments.

  • Software architecture
  • Systems integration
  • APIs & interoperability
  • Reliability & security
  • Traceability & audit trail

Payment Systems & Financial Standards

02

Integrating payment systems and financial flows structured by ISO 20022 and its associated standards.

  • ISO 20022
  • SWIFT / MX
  • Payment flows
  • Financial systems integration
  • Transaction processing

Accounting & Ledger Systems

03

Connecting operational flows to accounting systems: data consistency, reconciliation, reference data.

  • Accounting systems
  • Ledger integration
  • Reconciliation flows
  • Referential data
  • Data consistency

Digitisation, ECM & Archiving

04

Document digitisation, ECM and electronic archiving, with the applicable ISO, NF and retention standards.

  • Document digitisation
  • GED / ECM
  • Electronic archiving
  • ISO & NF standards
  • Retention & integrity

AI & Intelligent Systems

05

Modelling, predicting, learning, and knowing when AI is the wrong answer.

  • Machine Learning
  • NLP
  • Deep Learning
  • LLMs
  • AI-assisted development

Modern Development

06

Automating the repetitive so effort goes where judgement is required.

  • Automation
  • Developer tooling
  • Modern UI
  • Data-driven apps
  • Rapid prototyping

04 / ZOOM INTO THE SYSTEM

An application isn't a screen. It's a stack of decisions.

Pick a layer to see what it actually contains, and what it guarantees to the layers next to it.

USER · INTENT

Everything starts with an intent. The easiest layer to draw and the easiest one to betray.

TaskContextTrust

05 / SELECTED WORK

Five projects. Problem first, technology second.

PROJECT 01

Payment Systems & Financial Integration

Designing and integrating reliable payment systems in a regulated environment.

JavaSpring BootOracleISO 20022SWIFT / MXREST APIs
◦ Confidential: details cannot be shared

CHALLENGE

Making payment systems, reference data and accounting systems work together around demanding financial standards, in an environment where a flow error turns directly into an accounting discrepancy, not just a bug.

APPROACH

A business reading of the payment flow before any technical decision: flow mapping, compliance controls, integration points, accounting posting, end-to-end traceability. Every stage isolated, observable and replayable.

OUTCOME

Active contribution to the ISO 20022 compliance of a critical payment system: flow mapping, integration of several regulatory control points over multiple months. Technical details remain confidential, demonstrable at interview.

PROJECT 02

Enterprise Document Systems · Digitisation, ECM & Archiving

Making a digital document as trustworthy as an accounting entry.

GED / ECMDigitisationElectronic archivingISO & NFMetadata
◦ Confidential: details cannot be shared

CHALLENGE

A digitised document only has value if its integrity, lifecycle and retention can be demonstrated, a matter of standards as much as architecture.

APPROACH

The full chain: digitisation, indexing and metadata, ECM, then electronic archiving aligned with applicable ISO and NF standards and retention frameworks. Traceability and retention treated as design requirements, not options.

OUTCOME

Digitisation and electronic archiving chain deployed across multiple document types and internal departments, compliant with applicable ISO/NF frameworks. Volume and details remain confidential, demonstrable at interview.

PROJECT 03

Intelligent Battery Health Estimation

Predicting a cell's end of life before the cell reveals it.

PythonScikit-learnXGBoostGenetic AlgorithmLaTeX
GitHub ↗

CHALLENGE

Estimating the state of health of Li-ion batteries from early charge cycles, on the Toyota Research Institute dataset (124 cells tested to failure).

APPROACH

Per-cycle feature extraction, then an ensemble model (Random Forest, XGBoost, linear regression) whose weights are optimised by a genetic algorithm rather than hand-tuned.

OUTCOME

Final RMSE of 0.0114, 88% lower than the same unweighted ensemble. Master's thesis.

PROJECT 04

Pulse · Blood Donation Platform

Connecting an urgent blood need to a compatible nearby donor, before the stock turns critical.

FlutterDartFirebaseFigma
GitHub ↗

CHALLENGE

Blood donation rarely lacks donors: it lacks coordination between need, stock and proximity at the right moment.

APPROACH

Flutter/Firebase mobile app: a map of nearby donation centres, an eligibility questionnaire before donating, and appointment booking saved to a database.

OUTCOME

Selected for the Google Solution Challenge 2023, built by a team of 4 with 40+ public commits (full history on GitHub).

PROJECT 05

Conversational AI · BERT × GPT-2

Two models, one conversation, no script.

PythonPyTorchBERTGPT-2
GitHub ↗

CHALLENGE

Making two language models hold a coherent dialogue, well before LLMs became a commodity.

APPROACH

Turn-by-turn exchange pipeline: BERT for understanding and selection, GPT-2 for generation, with context-drift control.

OUTCOME

The basis of my DevFest NLP talk (2023) and of how I read LLMs today.

06 / EVOLUTION

Six years moving up the stack, from language to system.

NOW

PAYMENTS × DOCUMENTS × AI

ISO 20022 certification, AI as an engineering tool. Same method, wider scope.

Inside that evolution, AI isn't a destination but one more engineering tool: it speeds things up without replacing the understanding of the problem or the discipline of architecture.

AI→ Automation→ Code→ Product

07 / ARCHITECTURE → LOGIC → CODE

Code is the last step, not the first.

FUNCTION
INPUT
OUTPUT
LOGIC
01receive(payment)controlled intake
02validate(payment)business & compliance rules
03enrich(referential)reference data
04postToLedger(entry)accounting entry
05archive(record)traceability & retention
class TransferService {

    Result process(PaymentMessage msg) {
        var payment  = intake.receive(msg);
        var checked  = validator.validate(payment);
        var enriched = referential.enrich(checked);

        return repository.commit(enriched);
    }
}

The same intent at three zoom levels. That is the order I work in: structure, then flow, then implementation.

08 / BEYOND CODE

Software isn't written alone, and not in silence.

Share

DevFest speaker on AI and NLP, Python trainer at GDSC, computer science teacher. Explaining clarifies.

  • DevFest 2022 & 2023
  • GDSC Python Track
  • Accredited teaching

Explore

AI, science, philosophy, chess. Different boards, same requirement: understand the structure before you play.

  • Applied research
  • Philosophy
  • Chess

Build

Side projects, experiments, automation. The best way to understand a system is still to build one.

  • Side projects
  • Automation
  • Rapid prototyping

09 / CREDENTIALS

EDUCATION

Master's · Intelligent Computer Systems
Mouloud Mammeri University
2022-2024
Bachelor's · Computer Systems
Mouloud Mammeri University
2019-2022

LANGUAGES

FrenchAdvanced
EnglishAdvanced
ArabicAdvanced
KabyleNative

CERTIFICATIONS

ISO 20022 Directly tied to current role2026 · 01
Banking Intelligence Consulting (BIC)
SWIFT Payments · SWIFTNet · Financial messaging
Introduction to Software Engineering2025 · 09
IBM
Software engineering · Agile · Architecture
Verify ↗
Python for Everybody Specialization2023 · 08
Coursera
Python · Data structures · Web data · Databases
Verify ↗
Capstone: Retrieving, Processing & Visualizing Data2023 · 08
Coursera
Programming · Visualisation · Data analysis
Verify ↗
Using Databases with Python2023 · 08
Coursera
Relational databases · SQL · SQLite
Verify ↗
Using Python to Access Web Data2023 · 08
Coursera
Web scraping · APIs · Content extraction
Verify ↗
Fundamentals of Deep Learning2022 · 11
NVIDIA
Neural networks · CNNs · Image processing
Verify ↗
Cyber Security Operations2022 · 09
ICT Towers
Cybersecurity · Security operations · SOC
Google Cloud Essentials2022 · 09
Google Cloud Skills Boost
Cloud computing · GCP · Infrastructure
Verify ↗
Python Data Structures2022 · 08
Coursera
Data structures · Python · Algorithms
Verify ↗
Introduction to Artificial Intelligence (AI)2022 · 05
IBM
AI concepts · Impact · Ethics · Applications
Verify ↗
Programming for Everybody (Python)2022 · 02
Coursera
Python · Programming · Fundamentals
Verify ↗

10 / PEER FEEDBACK

"Working together across various technical projects, he's a solid pillar."
Hanine Bouchiba
Software Engineer · Backend & Distributed Systems
LinkedIn ↗
"A very brilliant and serious person, truly driven, always active and ready to solve any IT problem."
Arezki Kahlouche
Full-Stack Developer · AI & ML Enthusiast
LinkedIn ↗
"This person is the most creative I know in my life."
Yanis Douki
Developer
LinkedIn ↗
"Hard worker, always invested in the club's projects, with exemplary rigor."
Aomar Khelifi
MSc AI · NLP & Computer Vision research
LinkedIn ↗

11 / PARTNERS & INVESTORS

Exploring Algeria for investment or partnerships?

Based in Algiers, I know the local tech, banking and entrepreneurial ecosystem, and can be your point of contact on the ground.

NETWORKING & INTRODUCTIONS

Local networking

Looking to meet the right people in Algeria's tech, finance or entrepreneurship scene? I facilitate introductions and help you navigate the local ecosystem.

LIAISON & REPRESENTATION

Local representative

For companies entering or investing in the Algerian market, I can act as your local point of contact: coordination, follow-up, and an interface with people on the ground.

Let's talk about your project

What's next?

I'm always interested in complex problems worth solving: critical systems, integration, automation, applied AI.

Let's build something meaningful.

3 YEARS IN REGULATED SYSTEMS · ISO 20022 CERTIFIED · 5 PROJECTS SHIPPED

ResumeDownload PDF ↓ LinkedInsalim-mabed ↗ GitHubSalimMabed ↗ Emailcontact@salimmabed.com
LocationAlgiers, DZ · UTC+1