Ermal Belul portrait
ACTIVE — CIFRE PhD RESEARCH · 2023–2026

> Ermal
Belul

Operations Research Scientist & Systems Architect

Systems engineer and Operations Research scientist (PhD) transforming complex, noisy data into strategic, decision-ready intelligence. I bridge academic mathematical rigor and industrial scalability by designing algorithmic systems that improve decisions, reduce risk, and optimize performance under real-world physical constraints.

LinkedIn GitHub Email CV
//about

I am an Operations Research and Applied Scientist with deep expertise in algorithm design, software development, and mathematical modeling. My work focuses on translating complex methodologies such as Mixed-Integer Linear Programming (MILP), constraint programming, heuristics, and Machine Learning into robust, scalable software architectures.

Specializing in the strategic optimization of Supply Chain and intralogistics environments (WMS, TMS, OMS, ASRS), I architect distributed solutions capable of orchestrating complex physical constraints. My models have successfully reduced operational compute bottlenecks by up to 21% in massive-scale environments exceeding 20,000 nodes.

Beyond technical development, I lead comprehensive digitalization projects, driving everything from business process analysis to change management. I build decision-support systems that secure ROI, reduce risks, and streamline daily warehouse operations without causing physical disruption. Bilingual English / French.

//currently working on
ACTIVE PROJECT

PhD CIFRE — Operations Research for Intralogistics

UTC × Savoye

Engineered a set-variable Mixed-Integer Linear Programming (MILP) model for an automated warehouse dataset of 21,877 SKUs. Shifted optimization criteria from travel distance to station-visit minimization (CSLAP), structurally reducing daily mechanical stops by 13.7% (2,198 fewer visits/day) and recovering 9 full operating days per quarter.

MILPHexaly / CPLEXStochastic & Robust Opt.SLAP / CSLAP
ACTIVE PROJECT

MSLAP & Dynamic Slot Merging

Savoye

Designed and developed a 3-stage flow-based decomposition software framework to solve the novel Mergeable Storage Location Assignment Problem (MSLAP). Successfully bypassed branch-and-bound combinatorial explosions to activate 278 dynamic slot merges, cutting total operational costs (picking and replenishment) by 21% across a 3,600-location grid.

Network FlowHeuristicsBipartite MatchingAlgorithm Design
//research focus

Operations Research under Uncertainty

Stochastic and robust MILP formulations designed to make intralogistics decisions resilient to demand volatility and process variability.

Practical Algorithmic Scalability

Formulating mathematically bounded "limited-reassignment" algorithms. Proved that maximum throughput gains can be captured by relocating merely 2.2% of an inventory catalog (500 SKUs), ensuring zero disruption to continuous daily fulfillment operations.

Decision Systems & AI Integration

Architecting AI-assisted reporting ecosystems integrating LLM APIs and structured RAG foundations over large-scale datasets to automate executive-level decision support and operational tracking.

//experience
  1. Dec 2023 — PresentDijon, FR

    R&D Operations Research Scientist (Industrial PhD Candidate)

    Savoye
    • Algorithm Engineering & Optimization: Engineered a set-variable Mixed-Integer Linear Programming (MILP) model in Python for an automated warehouse dataset of 21,877 SKUs. Shifted optimization criteria to zone-visit minimization, reducing daily mechanical stops by 13.7% (2,198 fewer visits/day) and recovering 9 full operating days per quarter.
    • Large-Scale System Design: Designed and developed a 3-stage flow-based decomposition software framework to solve the Mergeable Storage Location Assignment Problem (MSLAP). Bypassed branch-and-bound combinatorial explosions to activate 278 dynamic slot merges, cutting total operational costs by 21% across a 3,600-location grid.
    • Production-Ready Deployments: Formulated and implemented a mathematically bounded "limited-reassignment" algorithm that captured maximum throughput gains by relocating merely 2.2% of the inventory catalog (500 SKUs), ensuring zero disruption to continuous daily fulfillment operations.
    • Technical Leadership: Directed and mentored an R&D student software task force to build a hybrid layout heuristic. Delivered a production-ready strategy that reduced daily station transitions by 8% (saving 5.4 working days per quarter) while maintaining strict workload equilibrium.
  2. Feb 2023 — PresentRemote

    Data Manager & Decision Systems Architect

    Upwork
    • Architected robust and scalable cloud data pipelines on GCP, BigQuery, and Databricks (Azure) for global SaaS platforms, reducing data processing time by 40%.
    • Deployed AI-assisted reporting ecosystems and GenAI architectures (RAG systems via LLM APIs) to automate decision support, generating a 33% increase in operational efficiency.
    • Integrated DBT into client infrastructures to modularize analytical logic, improving data traceability and reducing reporting errors by 80%.
  3. Feb 2023 — Nov 2023Dijon, FR

    AI & Operations Research Intern

    Savoye
    • Designed and deployed network flow algorithms to solve the Storage Location Assignment Problem (SLAP) using historical WMS data, simulating a 25% reduction in total operational picking and replenishment costs.
    • Engineered statistical models to evaluate inventory reorganization scenarios, proving that a targeted reallocation of just 20–40% of critical SKUs achieves near-optimal throughput while minimizing physical warehouse disruption.
  4. Jul 2024 — PresentCompiègne, FR

    Teaching Assistant (Operations Research & Python)

    Université de Technologie de Compiègne
    • Instructed engineering classes on foundational Operations Research literature, focusing heavily on Graph Theory and combinatorial optimization.
    • Designed and taught Python programming labs that translated theoretical mathematical logic and graph algorithms into executable software code.
    • Bridged the gap between academic theory and software engineering by teaching applied data structures and algorithmic problem-solving.
  5. May 2021 — Aug 2022Tirana, AL

    Data Delivery Engineer & Project Manager

    Engage3
    • Managed the delivery of complex data pipelines and market analyses utilizing PostgreSQL, JavaScript, and Python.
    • Engineered automated web extraction agents to ingest and transform competitive retail datasets into actionable reporting metrics.
    • Supported teams in managing change through local training and operational support for new employees.
//education
Dec 2023 — Present

Ph.D. in Operations Research, AI & Computer Science

Université de Technologie de Compiègne (UTC) · CIFRE w/ Savoye
Sep 2022 — Sep 2023

M.Sc. Complex Systems, Machine Learning & Optimization

Université de Technologie de Compiègne (UTC)
Sep 2021 — Jul 2023

M.Sc. Software Engineering

Polytechnic University of Tirana
Sep 2018 — Jul 2021

B.Sc. Electronic Engineering

Polytechnic University of Tirana
//selected publications
  1. 2026·Société Française de Recherche Opérationnelle et d'Aide à la Décision (ROADEF)

    Storage Location Assignment with Mergeable Locations

    E. Belul, D. Nace, A. Jouglet, M. Bouznif
  2. 2025·20th Annual System of Systems Engineering Conference (SOSE)

    Optimizing Automated Warehouse Operations: A Novel Approach to the Correlated Storage Location Assignment Problem (CSLAP)

    E. Belul, M. Bouznif, D. Nace, A. Jouglet
  3. 2024·Société Française de Recherche Opérationnelle et d'Aide à la Décision (ROADEF)

    Strategic Solutions for Warehouse Optimization: Tackling the Storage Location Assignment Problem

    E. Belul, D. Nace, A. Jouglet, M. Bouznif
  4. 2023·2nd International Conference on Information Technologies and Educational Engineering (ICITEE)

    Optimizing Warehouse Storage: The Location Assignment Problem

    E. Belul, D. Nace, A. Jouglet, M. Bouznif
//skills & tools
Operations Research
MILP / CPCPLEX / Hexaly / GurobiStochastic ModelingRobust OptimizationDecision under UncertaintyDynamic ProgrammingAdvanced Decomposition MethodsSimulationHeuristics & MetaheuristicsNetwork Flow TheoryCombinatorial Optimization
Data Science & AI
Predictive ModelingStatistical ModelingCausal ReasoningTime-series AnalysisLLM Integration / RAGTensorFlowPyTorchscikit-learnXGBoostPandas / NumPyHugging Face
Engineering Stack
PythonC / C++R (Tidyverse)SQL / PostgreSQLJavaScriptMATLAB
Cloud & Infrastructure
GCP / BigQueryDatabricks (Azure)dbtData PipelinesSystem Architecture
Supply Chain & Logistics
WMS / WCS / OMSASRS EnvironmentsEnd-to-End PlanningOmnichannel StrategyProcess MappingManagementTeam Leading
Languages
French — B2English — BilingualAlbanian — NativeItalian — A2
//contact

Let's build decision systems that ship.

Open to research collaborations, advisory roles, and industrial projects bridging operations research, AI, and intralogistics.

LinkedIn GitHub ermalbeluli10@gmail.com ermal.belul@hds.utc.fr ermal.belul@fti.edu.al