Développeur fullstack × Professeur × Creative Engineering

Alexandre-Benjamin
Zérah

Créativité | Connaissances — Frontend, Backend, Algorithmes, Systemes.

01

A propos

Abstract

Ingénieur en logiciels à l’intersection de l'informatique appliquée et de l’architecture des systèmes. Mon travail repose sur le raisonnement formel, la conception algorithmique et une vision structurelle du code.

Je conçois des architectures backend et des systèmes distribués en les abordant comme des objets structurés : identification des invariants, maîtrise de la complexité, analyse des performances, cohérence sous contrainte et capacité d’évolution. La scalabilité n’est pas un ajout, elle est pensée dès la conception.

En parallèle des systèmes de production, j’explore le creative coding et les processus génératifs. La programmation devient alors un médium : un espace où la rigueur mathématique rencontre la forme, l’image et la perception.

02

Travail de recherche

01

Conception algorithmique

Conception formelle d’algorithmes efficaces avec analyse rigoureuse de la complexité. Travail centré sur la théorie des graphes, l’optimisation et les structures combinatoires.

02

Architecture backend

Conception de systèmes scalables et tolérants aux pannes. Attention portée à la qualité des abstractions, à l’intégrité des flux de données et à la clarté des interfaces API.

03

Systèmes distribués

Mise en œuvre de protocoles de consensus, modèles de cohérence éventuelle et conception d’architectures multi-nœuds résilientes.

04

Pensée computationnelle

Application du raisonnement mathématique pour décomposer des problèmes complexes en unités de calcul maîtrisables et vérifiables.

05

Creative coding & systèmes génératifs

Génération algorithmique de formes visuelles et structurelles. Design paramétrique, géométrie procédurale et exploration esthétique par le calcul.

06

Intelligence artificielle appliquée

Intégration de modèles d’apprentissage automatique dans des systèmes de production. Optimisation de l’inférence, architecture de serving et maîtrise des contraintes opérationnelles.

03

Selected Work

Distributed Task Orchestrator

Go, gRPC, Redis, Kubernetes

Problem

Coordinate heterogeneous compute tasks across unreliable nodes with minimal latency overhead.

Approach

Implemented a DAG-based scheduler with priority queues and circuit-breaker patterns for fault isolation.

Outcome

Achieved sub-50ms scheduling latency at 10K concurrent tasks with 99.97% delivery guarantee.

Real-Time Anomaly Detection Pipeline

Rust, Apache Kafka, ClickHouse, Python

Problem

Detect statistical anomalies in high-volume streaming data with bounded memory consumption.

Approach

Designed a sliding-window algorithm using exponential histograms and approximate quantile sketches.

Outcome

Processing 2M events/sec with p99 detection latency under 200ms and 0.3% false positive rate.

Generative Topology Visualizer

TypeScript, WebGL, Three.js, WASM

Problem

Render interactive visualizations of topological spaces for mathematical education.

Approach

Built a parametric rendering engine mapping abstract simplicial complexes to navigable 3D meshes.

Outcome

Used by 3 university departments. Renders complexes with up to 50K simplices at 60fps.

API Gateway with Adaptive Rate Limiting

Go, PostgreSQL, Envoy, Prometheus

Problem

Design a gateway that adapts rate limits dynamically based on system load and client behavior patterns.

Approach

Token bucket algorithm enhanced with Bayesian inference for predictive load estimation.

Outcome

Reduced cascading failures by 94% while maintaining throughput under 5% overhead.

04

Technical Stack

L0

Languages

GoRustTypeScriptPythonSQLOCaml
L1

Systems

PostgreSQLRedisKafkaClickHouseElasticsearchgRPC
L2

Infrastructure

KubernetesDockerTerraformAWSCloudflarePrometheus
L3

Tooling

GitNixNeovimCI/CDOpenTelemetryGrafana
05

Mathematical Thinking

Mathematics is not a tool applied after the fact — it is the substrate upon which all architectural decisions are formed.

Invariant-First Design

Every system begins with the identification of invariants — properties that must hold regardless of state transitions. This mathematical discipline ensures correctness by construction rather than by testing.

λ

Compositional Architecture

Systems are composed from small, well-defined functions with clear type signatures. Inspired by category theory, each module is a morphism in a larger architectural diagram.

Constraint Propagation

Complex problems are modeled as constraint satisfaction instances. By propagating constraints early, the solution space is pruned to tractable dimensions before implementation begins.

Asymptotic Reasoning

Performance is not measured in benchmarks alone but analyzed through asymptotic complexity. Every data structure choice is a theorem about expected workload behavior at scale.

06

Publications & Notes

2025

On the Convergence Properties of Adaptive Rate-Limiting Algorithms

Technical Report
2024

Compositional Patterns in Distributed Event-Driven Architectures

Systems Design Notes
2024

A Category-Theoretic Approach to API Surface Design

Blog / Research Note
2023

Efficient Approximate Quantile Estimation for Streaming Data

Internal Publication
2023

Notes on Parametric Curve Generation for Real-Time Rendering

Creative Coding Journal
07

Education

M

MUSIC - MSc Applied Mathematics & Computer Science

Paris, France2023 — 2025

Master of Science in Applied Mathematics and Computer Science. Research-oriented program at the intersection of mathematical modeling, optimization, and software engineering.

Key coursework

  • Advanced Optimization & Operations Research
  • Stochastic Processes & Probabilistic Modeling
  • Machine Learning Theory & Statistical Learning
  • Distributed Systems & High-Performance Computing
  • Research Project: Convergence analysis of adaptive algorithms
M

MUSIC - BSc Mathematics & Computer Science

Paris, France2020 — 2023

Bachelor of Science combining pure mathematics foundations with applied computer science. Strong emphasis on formal reasoning and algorithmic thinking.

Key coursework

  • Linear Algebra & Abstract Algebra
  • Real & Complex Analysis
  • Probability Theory & Statistics
  • Data Structures & Algorithm Design
  • Numerical Methods & Scientific Computing
  • Database Systems & Software Architecture
P

Preparatory Classes — MPSI / MP

France2018 — 2020

Intensive two-year program in mathematics, physics, and computer science. Highly selective French preparatory classes for engineering schools.

Key coursework

  • Advanced Mathematics (Algebra, Analysis, Topology)
  • Theoretical Physics (Mechanics, Thermodynamics, Electromagnetism)
  • Computer Science (Algorithms, Formal Logic, OCaml)
  • Oral & Written Competitive Examinations
08

Contact

Selective inquiries
welcome.

For research collaborations, technical consulting, or projects that require mathematical rigor in their engineering — a conversation can begin here.