Skip to content
View foersben's full-sized avatar
:electron:
Focusing
:electron:
Focusing

Block or report foersben

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
foersben/README.md

👋 Hi, I'm Benjamin.

I'm a Ph.D. Researcher (unemployed) actively transitioning into roles as a Generative AI Developer, AI Architect, and Data Scientist. My work bridges the gap between mathematically rigorous complex systems research and high-performance, constraint-driven software engineering.

🌟 Flagship Project

  • PHIDS: A deterministic ecological simulation framework developed independently with academic and publication support from Uni Jena (PD Dr.-Ing. habil. Thomas Hinze). Engineered for uncompromised performance, it utilizes a strict Entity-Component-System (ECS) architecture, Numba JIT compilation, and O(1) spatial hashing to guarantee invariant execution. It features memory-decoupled Zarr/Polars telemetry pipelines and a natively integrated Model Context Protocol (MCP) server allowing external LLMs to read runtime snapshots and execute self-evolving experiments.

🚀 Advanced Engineering & AI Projects

  • CodaCite: A proof-of-concept, local-first GraphRAG document intelligence engine. Built with SurrealDB 3.0, DeepSeek-R1, and LangGraph on a modular Vertical Slice Architecture, it achieves character-level citation accuracy and multi-hop reasoning with zero data leakage.
  • Project Janus: Synapse Edge 2026: An exercise in extreme Constraint Engineering and hardware-software co-design. It deploys a Dual-Engine inference stack (llama.cpp 70B/7B) and virtualized cognitive memory (Letta) on a consumer-grade 20L footprint, utilizing a Rust-native Rig framework and zero-trust bare-metal topologies (Proxmox, Talos Linux) (2.5th revision this year and probably at least partially outdated again).
  • Industrial Component Anomaly Detection: An end-to-end computer vision pipeline evaluating MVTec datasets. Built on a strict Vertical Slice Architecture using Pixi, it deploys deep learning CNNs via supervised and unsupervised approaches to perform binary defect isolation followed by multi-class classification for highly localized flaws.

🔬 Academic & Research Background

  • Research Focus: Realising heterogeneous security landscapes for distributed collaborative security in large-scale static Wireless Sensor Networks (WSNs). My work provides optimal trade-off analyses of resource utilization, longevity, and robustness—aiming to broaden threat coverage while strictly minimizing energy expenditure.
  • Authorship: Lead Author on 6+ peer-reviewed publications and three DFG project proposals on WSN security; Contributing Author on 2 additional papers.

Pinned Loading

  1. PHIDS PHIDS Public

    Plant-Herbivore Interaction & Defense Simulator (PHIDS): A visual discrete event-based simulator modelling ecological dynamics between plants and herbivores on a spatial grid. It tracks energy, pop…

    Python 1

  2. janus-synapse-edge janus-synapse-edge Public

    Project Janus: Synapse Edge. A dual-engine autonomous cluster (llama.cpp) utilizing NVFP4 on an RTX 5070 Ti. Features virtualized cognitive memory (Letta), self-evolving idle-loops (GEPA), and mult…

    Python

  3. CodaCite CodaCite Public

    CodaCite: CPU-only local-LLM KG-RAG-based Document Intelligence. Local-first semantic search and knowledge graph reasoning with verifiable, multi-hop citations.

    Python 1

  4. where-next where-next Public

    Personalised country recommendation engine — MILP + Gemini AI + polyglot data backend

    Python 1

  5. lambda_precision_udg_generator lambda_precision_udg_generator Public

    A toolkit for generating λ -precision UDGs, primarily designed for modelling large-scale static WSNs. It features a suite of solver-agnostic MILPs. Whether you are minimising network variance or ba…

    Python

  6. industrial-component-anomaly-detection industrial-component-anomaly-detection Public

    End-to-end computer vision pipeline for industrial anomaly detection using MVTec datasets. Built on a strict Vertical Slice Architecture with Pixi, it deploys deep learning arch (CNNs) via supervis…

    Jupyter Notebook