Skip to content
LymperisPerakis
Open to interesting conversations

LymperisPerakis.

→AI Engineering Manager with a background in ML systems and backend architecture. I lead cross-functional teams shipping production AI — currently an AI agent that supports electronics design workflows.

Download CVGet in touchLinkedIn
Lymperis Perakis portrait
Currently
AI Engineering Manager at CELUS
01About

A short note on me.

AI Engineering Manager with a strong background in machine learning systems and backend architecture. Experienced in leading cross-functional teams and deploying production AI solutions, including an AI agent supporting electronics design workflows. Focused on translating business needs into scalable technical systems and delivering measurable product impact.

What I focus on

Pragmatic AI systems, healthy team rituals, and translating product intent into shippable engineering work.

How I work

Calmly, with curiosity. I write things down, prefer small reversible bets, and protect my team's focus.

Currently working with.

Toolkit
AI
  • Machine Learning Systems
  • AI Agents
  • LLM Applications
  • MLOps
Engineering
  • Backend Architecture
  • Data Pipelines & ETL
  • Cloud-native Systems
  • Python
Leadership
  • Engineering Management
  • Hiring & Team Building
  • Technical Strategy
  • Stakeholder Management
02Work

Roles & trajectory.

From research to leadership — the path that brought me to managing engineering teams.

  1. Oct 2023 — PresentLeadership

    AI Engineering Manager

    CELUS·Munich, DE
    • Lead an AI engineering team developing intelligent systems supporting electronics.
    • Architect and deploy an AI agent assisting users in electronics design.
    • Translate product and business requirements into scalable AI solutions.
    • Establish engineering practices for AI development, deployment, and monitoring.
  2. Jan 2021 — Sep 2023Engineering

    Tech Lead

    CELUS·Munich, DE
    • Led backend architecture development and mentored engineers across multiple teams.
    • Developed APIs enabling data integration across internal services and external systems.
    • Contributed to DevOps practices including CI/CD pipelines, containerization, and cloud infrastructure.
  3. Jun 2019 — Dec 2020Research

    ML Researcher

    CELUS·Munich, DE
    • Implemented NLP for extracting structured data from electronic component datasheets.
  4. Sep 2017 — Mar 2018Operations

    Regional Project Manager

    Artemis ITS·Cologne, DE
    • Managed and supervised three teams of 13 in FTTx projects, ensuring efficient execution and delivery.
    • Implemented a comprehensive documentation process for capturing FTTx project tracks, enhancing transparency and accountability.
03Education

Where I studied.

Technical University of Munich logo

Technical University of Munich

Munich, DE·2017 — 2022
M.Sc. in Electrical Engineering & Information Technology
  • Focus: Automation & Robotics.
  • Thesis: Analyzing illustrations and figures in electronics datasheets with methods of computer vision and machine learning.
Technical University of Munich logo

Technical University of Munich

Munich, DE·2017 — 2020
M.Sc. in Management
  • Focus: Innovation & Entrepreneurship.
  • Thesis: The use of machine learning techniques to analyze textual data for business.
Technical University of Munich logo

Technical University of Munich

Munich, DE·2014 — 2017
B.Sc. in Electrical Engineering & Information Technology
  • Core modules: Automation, Robotics, Electronics, Computer Science.
  • Thesis: Calculation of the domain wall velocity using micromagnetic simulations in films with perpendicular anisotropy.
Deutsche Schule Athen (DSA) logo

Deutsche Schule Athen (DSA)

Athens, GR·2004 — 2014
Abitur · Panhellenic Examinations
  • One of the oldest German international schools in Athens, with strong academic standards and a multicultural environment.
  • Completed both the German Abitur and the Greek Panhellenic Examinations.
04Projects

Selected work.

A growing list — both from work and from things I build on the side.

Shipped01 / 01

Human or animal?

The first AI project I worked on — a tiny CNN that decides whether a scanned figure is a human or an animal. Originally MATLAB; rewrote the training in PyTorch and exported to ONNX so it runs entirely in your browser, with a layer-by-layer walkthrough of the network.

PyTorchONNXComputer visionIn-browser
Open →
05Research

Things I've written.

23INFORMATIK 2023

Classifying figures and illustrations in electronics datasheets: A comparative evaluation of recent computer vision models on a custom collection of 4000 technical documents

AuthorsLymperis Perakis, Julian Balling, Frank Binder, Gerhard Heyer, Franz Kreupl

A comparative evaluation of several recent object-detection models applied to technical document analysis and graphics recognition. YOLOv7-D6 was found to be the most accurate model for classifying figures in electronics datasheets.

Digital libraryRead PDF