Available for consulting & engineering projects
Mike Kertser

Mike Kertser

Physicist • Algorithm Developer • R&D Lead

I build physics-driven simulations and ML/DS solutions that ship—from optical/thermal/flow modeling to production-ready data pipelines and decision tools for real systems.

Python • MATLAB • C/C++ ML/DS • Deep Learning CFD • FEA • Multiphysics ANSYS • COMSOL • OpenFOAM Optimization & Decision Tools

Currently: Head of Physics at Atlantium Technologies (UV & UV-LED disinfection systems).
Operating as Alpha-Numerical for select consulting projects.

What you get:

A validated model, clear report, or working tool—not slides about future work. Typical deliverables: simulation model with validation, design trade-off analysis, prediction pipeline, or optimizer/dashboard ready for use.

Quick Facts

  • 20+ years in Physics & Engineering
  • 5+ years in Data Science and Machine Learning
  • 20+ patents (UV systems, medical devices)
  • M.Sc. Physics, Tel-Aviv University
  • B.Sc. Physics & CS, Ben-Gurion University

Domains

UV disinfection systems • Medical sensors & respiratory devices • Optical systems • Thermal & fluid dynamics • Data Science & ML applications

Languages

Russian (native) • English (fluent) • Hebrew (fluent)

Engagement

Remote-first • Based in Israel (GMT+2) • NDA-friendly • Quick turnaround available

20+
Years Experience
20+
Patents Filed
3
Core Domains
15+
ML/DS Projects

What I Do

End-to-end solutions combining physics, algorithms, and machine learning

⚙️

Simulation & Modeling

Physics-based simulations for design optimization and validation.

  • CFD (SW, OpenFOAM, ANSYS Fluent)
  • FEA & Thermal Analysis
  • Optical ray tracing (TracePro, APEX/ASAP)
  • Multiphysics (COMSOL)
  • Parametric studies & sensitivity analysis
🤖

ML / Data Science

Production-ready ML solutions for real-world problems.

  • Classical ML (scikit-learn, XGBoost)
  • Deep Learning (PyTorch, TensorFlow)
  • NLP, Computer Vision, Time Series
  • Feature engineering & model deployment
  • Statistical modeling & visualization
🛠️

Tools & Prototypes

Custom tools that bridge theory and practice.

  • Optimization calculators & decision tools
  • Interactive dashboards
  • Automated analysis pipelines
  • Rapid 3D prototyping
  • System integration & deployment

Experience

20+ years bridging experimental physics and practical engineering

Atlantium Technologies — Head of Physics

2020 – Present
  • Leading multidisciplinary R&D for UV and UV-LED disinfection systems
  • System physics, numerical calculations, multi-domain simulations
  • Optical/thermal/mechanical/hydraulic design and optimization
  • ML/DS projects for process optimization and system monitoring
  • Managing research teams and coordinating cross-functional projects

Alpha-Numerical — Developer (self-employed)

2020 – Present
  • Physical simulations: CFD, Mechanical, Thermal, Optical modeling
  • Data Science & Machine Learning consulting projects
  • Algorithm development and optimization tools
  • Rapid prototyping and 3D printing solutions

Medtronic (Jerusalem) — Senior Physicist

2015 – 2020
  • Predictive engineering (FEA/CFD/thermal/optical) for medical devices
  • Optical sensors development and validation
  • Tech lead for capnography systems—concept to production delivery
  • Pneumatic research lab: multidisciplinary experimental work
  • Multiple patents in respiratory monitoring and oxygen delivery

Atlantium Technologies — Physicist at R&D

2006 – 2014
  • System design and engineering for UV-based hydro-optical systems
  • Experimental work in optics, acoustics, thermal, mechanics, hydrodynamics
  • Computational simulations and optimization algorithms
  • Company representative for UV technology

Earlier Roles

2003 – 2006

Intel (Kiryat-Gat) — Quality & Reliability Engineer
Tel-Aviv University — Research Assistant (High-pressure physics, XRD/XAS spectroscopy)
Full details and publication history available in CV

Skills & Expertise

Comprehensive technical toolkit spanning simulation, ML, and software engineering

🔬 Simulation & Physics

CFD & Flow Analysis OpenFOAM, ANSYS Fluent, SW Flow simulations
FEA & Mechanical ANSYS, SolidWorks Simulation, structural analysis
Thermal & Optical COMSOL, TracePro, RayViz, APEX/ASAP
3D Modeling SolidWorks, FreeCAD, rapid prototyping and 3D printing

🤖 ML & Data Science

Classical ML Clustering, Decision Trees, PCA, Boosting, Monte Carlo, Time-series
Deep Learning PyTorch, TensorFlow/Keras, NLP, CNN, RNN
Data Engineering Pandas, NumPy, SQL/NoSQL (MySQL/MongoDB)
Statistical Analysis and Visualization Matplotlib, Seaborn, Plotly, Tableau, Minitab

💻 Programming & Tools

Languages Python, MATLAB, C/C++, JavaScript, HTML/CSS
Python Stack scikit-learn, Pandas, NumPy, PyTorch, TensorFlow
Dev Tools Git, Docker
Computer Vision OpenCV, embedded systems (Raspberry Pi)

📊 Analysis & Methods

Statistical Analysis Design of Experiments, Six Sigma (Green Belt)
Optimization Linear programming, multi-objective optimization
Experimental Spectroscopy (XRD, XAS, Raman), lab instrumentation
Documentation Technical reports, patents, scientific publications

Featured Projects

Selected examples of simulation, ML, and engineering work

UV Systems Optimizer

Optimization • Web Tool • Multi-physics

Web-based calculator and optimization tool for UV system parametric sizing, combining optical ray-tracing with CFD simulations.

Challenge: Complex multi-parameter design space with coupled optical, hydraulic, and thermal variables
Solution: Linear programming optimization integrated with validated simulation models
Impact: Reduced design iteration time from weeks to hours

Keyboard Layout Switcher (DL)

NLP • Deep Learning • Open Source

Deep learning classifier for automatic keyboard layout switching (Russian-English-Hebrew) using positional embeddings and LSTM.

Tech Stack: Python, PyTorch, LSTM neural networks, positional embeddings
Repository: github.com/kertser/KeyboardSwitcher
Result: High-accuracy multilingual text classification

LLM-based Forum Assistant

NLP • Multi-modal • Cloud Infrastructure

Multi-modal conversational AI assistant based on large corpus of forum dialogues, with dialogue and image generation capabilities.

Architecture: GPT-2 → GPT-3.5-turbo pipeline with custom web scraper
Infrastructure: Oracle Cloud with XenForo Forum integration
Components: Server, client, and data pipeline automation

RAG Technical Document Assistant

NLP • RAG • Enterprise

Retrieval-augmented generation chatbot for technical documentation with multimodal capabilities.

Purpose: Intelligent search and Q&A for complex technical documents
Technology: Vector databases, semantic search, context-aware responses

Embedded Face Detector (RPi)

Computer Vision • Embedded ML

Lightweight face detection and classification system running on Raspberry Pi Zero W2 with custom camera.

Constraints: Limited compute, real-time processing, low power
Methods: Optimized CNN models, efficient preprocessing pipeline
Outcome: Deployable prototype with measurable accuracy/latency metrics

AI Job Hunter

AI Agents • LLM • Automation

Automated LinkedIn job discovery, scoring, and Easy-Apply pipeline powered by LLM-based resume matching, browser automation, and market-intelligence analytics.

Stack: Python, FastAPI, HTMX, Playwright, OpenAI, SQLite
Highlights: Embedding + LLM fit scoring, market-trend analysis, multi-user web GUI, scheduled runs, Docker deployment

TDG — Tactical Decision Game

Simulation • Multiplayer • LLM Agents

Web-based multiplayer tactical command-and-staff exercise platform with AI-controlled opposing forces, terrain intelligence, collaborative map drawing, and structured natural-language order parsing.

Stack: Python, FastAPI, PostgreSQL/PostGIS, Redis, Leaflet, llama.cpp, OpenAI
Highlights: Tick-based rules engine, terrain-aware A* pathfinding, LOS viewshed, doctrine-aware LLM prompting, EN/RU bilingual UI, session replay with AAR

SmartVoter

Data Science • Civic Analytics • NLP

Evidence-based political-match web app for Israeli parliamentary parties, combining real Knesset voting records, sponsored bills, party platforms, and LLM-assisted classification with an adaptive questionnaire.

Stack: Python, FastAPI, Next.js, PostgreSQL, Redis, OpenAI, Docker
Highlights: Adaptive convergence-driven questionnaire, Monte Carlo Knesset seat simulation, D'Hondt allocation, EN/HE/RU i18n, live poll refresh via web search

WatchBird

Computer Vision • Real-time ML • Edge

Real-time face-recognition and person-classification pipeline for embedded devices — distinguishing friendly, unknown, and armed subjects from a live camera stream.

Stack: Python, ONNX Runtime, DirectML/CUDA, SCRFD, ArcFace, YOLOv8, CLIP
Highlights: GPU-accelerated detection + tracking, PLDA scoring, CLIP-based soldier/armed/civilian classification, MJPEG streaming, state-machine identity confirmation

Medical Data-Science Study — Mitochondrial Glucose Metabolism

Data Science • Biomedical Research • Signal Analysis

Applied data-science research on glucose metabolism inside mitochondria, developing analysis pipelines and models to detect and characterise metabolic signals from experimental biomedical data.

Stack: Python, Jupyter, pandas, NumPy, SciPy, scikit-learn
Highlights: Exploratory data analysis, signal processing, feature engineering, statistical and ML modelling of mitochondrial metabolic activity
Context: Private R&D collaboration (repository non-public)

Portfolio Monte Carlo Simulator

Finance • Simulation • Web App

Local Dash web application for scenario-based portfolio accumulation analysis — downloads market data, computes risk statistics, and runs Monte Carlo projections with contribution schedules, fees, taxes, and rebalancing.

Stack: Python, Dash, Plotly, yfinance, pandas, NumPy, SciPy, Docker
Highlights: Five simulation models (historical/block bootstrap, parametric normal, Student-t, regime), model-comparison mode, rich Plotly visualisations, CSV/JSON export

Get in Touch

Ready to discuss your project? Multiple ways to reach me

Contact Information

📞
💻
📍
Location
Gedera, Israel (GMT+2)

Quick Start

To get started quickly, email me with three key pieces of information:

1. Your Problem What domain/challenge are you facing? (e.g., "Need CFD analysis for heat exchanger design")
2. Desired Output What deliverable do you need? (e.g., "Validated simulation model and optimization report")
3. Timeline & Constraints When do you need it? Any data/NDA/budget constraints?

💡 What to Expect

I typically respond within 24 hours with initial feasibility assessment and timeline estimate. Most projects range from 1-4 weeks depending on scope. NDA-friendly and remote-first workflow.