ML-Driven Automation for
Chemical & Materials

Boost production specifications and yield via AI/ML. Discover through closed-loop experimentation and machine learning.

Physics and ML background applied to chemistry's automation bottleneck. I build the systems that enable 10-100x faster experimental iteration.

Core Capabilities

Real-Time Monitoring & Control

  • Transform traditional analytical techniques into ML-optimized systems
  • Spectroscopy (UV/Vis, Raman, IR, fluorescence, etc.)
  • Mass spectrometry integration and control
  • Flow cytometry automation
  • Imaging systems (hyperspectral, computer vision, microscopy)
  • Computer vision and 3D sensing
  • Millisecond-scale feedback loops
  • Edge computing pipelines for real-time decisions

ML Optimization

  • Bayesian optimization for synthesis
  • Active learning frameworks
  • Multi-objective design
  • ML-steered nanoparticle synthesis
  • Automated parameter exploration

Hardware Integration

  • Microfluidic reactor systems
  • Flow chemistry automation
  • Sensor retrofitting & modernization
  • Legacy instrument cloud connectivity
  • Custom experiment platforms

My Strength: Cross-Domain Systems Transformation

I transform instruments, techniques, processes, procuts, and teams With techniques from systems thinking, programming, IoT, physics, and custom algorithm design, fast prototypoing. This cross-domain approach enables rapid innovation across different analytical methods and production environments.

ML & Algorithms

Leveraging open source tools across many fields, like a typical data scientist would

Custom ML Models

Well designed deep neural networks, Bayesian networks, and physics-inspired ML for unique challenges

Systems Thinking

End-to-end redesign from sensors to insights to decisions

IoT & Hardware

Connect, modernize, and control instruments for real-time feedback

Physics-Informed Design

Leverage first principles to build models that work with limited data

The methodology is transferable: understand the data generation process, redesign for ML optimization, build custom algorithms that extract maximum signal, deploy systems that enable fast iteration, facilitate and train teams to excute faster than anyone.

RESEARCH DIRECTIONS

Two Research Streams

1. Dynamic Chemistry Control

Finding and using optimal conditions to access reaction pathways and products that are yet unknown.

"Boston Dynamics for Chemistry"

Through mastery of Newtonian mechanics, Boston Dynamics robots access states that are hard for humans or animals to reach through natural training alone. Similarly, temporal control of reaction conditions allows access to chemical states that are hard to reach in classic production processes—optimizing them to be more efficient, safer, cheaper, and reduce waste.

Current Target Reducing toxic CTAB in gold nanoparticle synthesis through ML-guided temporal protocols
Key Innovation Real-time spectroscopy feedback enables discovery of optimal timing parameters

2. ML-Powered Nanoparticle Quality Control

Real-time quality control of nanoparticles using spectroscopy and single particle tracking—catching defects before use, not after.

The Problem for Nanoparticle Users

You receive batches from suppliers and need answers fast: Are they the right size? Correct shape? Acceptable quality? Traditional TEM/SEM takes days—by then, you've already committed them to your process.

Fast Incoming QC Characterize purchased nanoparticles in minutes—verify before use
Production Monitoring For synthesizers: catch bad batches during production, not after
ML-Powered Analysis Spectroscopy / particle tracking → instant size/shape/quality predictions

Whether you're using nanoparticles in your products or producing them at scale, this technology enables rapid QC, automated process control, and optimization cycles—transforming nanoparticle quality control from a data-poor bottleneck into a competitive advantage.

Collaboration Interests

I'm actively seeking:

  • → Chemistry partners with nanoparticle synthesis expertise
  • → Industry partners needing real-time QC solutions
  • → Materials companies wanting to reduce batch failures
  • → Academic groups interested in autonomous experimentation
  • → Investors/advisors familiar with deep tech commercialization

SpectroGroove

Your unified platform for spectroscopic data management, analysis, and knowledge extraction

The Problem

Spectroscopic data is scattered across instruments, spreadsheets, lab notebooks, and research papers. Extracting insights requires manually digging through files, converting formats, and cross-referencing multiple sources. Knowledge from papers stays locked in PDFs while your experimental data sits isolated in instrument software.

The Solution

SpectroGroove unifies your experimental spectroscopy data with knowledge from literature in one intelligent platform. Store, search, visualize, and analyze all your spectroscopic data while leveraging LLM-powered tools to extract protocols and insights from research papers - all in one place.

Key Features

📊 Data Visualization

Display spectrograms, browse data collections, visualize particle distributions and analysis results

🔍 Smart Search

Find your experimental data instantly - search across all your spectroscopy files and metadata

📁 Unified Storage

Centralized repository for all spectroscopic data - organize by project, sample, or experiment

🤖 LLM Integration

Extract protocols from papers, get analysis suggestions, automate routine tasks with AI assistance

📄 Paper Integration

Import data and methods from research papers - connect literature to your experiments

⚙️ Workflow Automation

Create automated analysis pipelines, execute shell commands, run custom scripts

📈 Particle Analysis

Calculate and visualize nanoparticle size/shape distributions from spectroscopic data

🔀 Multi-Project Support

Manage multiple parallel projects on one unified platform

📊 Spreadsheet Integration

Work with data in familiar spreadsheet formats, import/export seamlessly

Use Cases

Nanoparticle Synthesis Labs

Track UV/Vis spectra across batches, visualize particle size distributions, compare synthesis runs, extract methods from papers

Materials Research Groups

Centralize spectroscopic data from multiple instruments, search across projects, automate analysis workflows

Quality Control Teams

Store reference spectra, quickly find similar batches, automate pass/fail analysis, maintain audit trails

Academic Researchers

Build your spectroscopy database, extract protocols from literature, connect experimental data to published methods

Why SpectroGroove?

🔗 Connects Your Data with Literature

Stop switching between instrument software, spreadsheets, and PDFs. SpectroGroove brings experimental data and published protocols into one searchable system.

🤖 AI-Powered Knowledge Extraction

LLM integration helps you extract synthesis conditions from papers, suggest analysis approaches, and automate routine data processing tasks.

⚡ Built for Real Workflows

Designed for how spectroscopy actually works in labs - file uploads, shell commands, custom scripts, spreadsheets - not just pretty visualizations.

Interested in SpectroGroove?

Currently in active development. Contact me to discuss early access, custom integrations, or specific features for your lab's workflow.

Get in Touch

How I Can Help

My Approach: ML-Optimized Systems

I redesign your sensors, instruments, equipment, experiments, or production processes to maximize the value of the data they produce. Then I build custom ML models that extract maximum signal from that optimized data—giving you a significant competitive advantage.

Most AI advantage comes from two sources: unique, high-quality data (my ML-first hardware) and fast iteration cycles (automation + human-machine collaboration).

My workflows and tools smooth the collaboration between lab scientists, data scientists, and machines—removing barriers and accelerating discovery.

Strategic Consulting

Planning & technical strategy
  • ML-first hardware/experiment redesign feasibility
  • Competitive advantage assessment (unique data + fast iteration)
  • Real-time quality control system design (nanoparticles, materials)
  • Sensor selection and integration strategy
  • Data architecture design for experimental workflows
  • Technology due diligence for investors/acquirers
  • Experiment reduction strategy with ML
Deliverable: Reports, recommendations, technical roadmaps
Timeline: Days to weeks

Custom ML & AI Development

Maximum signal extraction from optimized data
  • Custom models designed for your ML-optimized hardware
  • Real-time nanoparticle QC (size/shape/quality from spectroscopy)
  • Deep learning for spectroscopy, mass spec, imaging data
  • Computer vision for microscopy, hyperspectral, visual QC
  • Time-series models for flow cytometry, real-time monitoring
  • LLM + human-in-the-loop workflows for lab operations
  • Experiment reduction systems with active learning
  • Custom labeling tools for domain experts
  • ML optimization pipelines (Bayesian, RL, evolutionary)
Deliverable: Production models trained on your unique data
Timeline: Weeks to months

Hardware & Sensing Systems

ML-optimized instrumentation & integration
  • Transform any analytical technique for maximum ML data value
  • Real-time QC systems for nanoparticle/material synthesis
  • Spectroscopy systems (UV/Vis, Raman, IR, fluorescence)
  • Mass spectrometry, flow cytometry integration
  • Imaging (hyperspectral cameras, computer vision, microscopy)
  • LiDAR and 3D sensing systems
  • Microfluidic/flow reactor design and control
  • Legacy equipment retrofit with modern sensing
  • Real-time monitoring with edge computing
  • Closed-loop experimentation platforms
Deliverable: ML-optimized hardware, control systems, prototypes
Timeline: Weeks to months

Data Infrastructure & Operations

Human-machine collaboration systems
  • Smooth lab scientist ↔ data scientist ↔ machine workflows
  • Remove barriers between researchers and their data
  • Organize data capturing flows (sensors → cloud)
  • Data pipeline architecture for fast iteration cycles
  • Custom tooling for non-programmers to access ML
  • Quality control & validation systems
  • Automated reporting and visualization dashboards
Deliverable: Production systems enabling fast iteration
Timeline: Weeks to months

Team Building & Leadership

Strategic technical leadership
  • Fractional CTO / Head of Data Science
  • Building and training data science teams
  • Technical roadmap and architecture decisions
  • Hiring, onboarding, and capability development
  • Bridge between research, engineering, and business
Deliverable: Ongoing leadership, team capability, strategic direction
Timeline: Months to years (ongoing)

Research Collaboration

Joint R&D partnerships
  • Joint development of novel synthesis methods
  • Co-authorship on publications
  • Shared IP development with clear terms
  • Grant proposal collaboration (SBIR, NSF, DOE)
  • Academic-industry partnership facilitation
Deliverable: Papers, patents, validated technologies
Timeline: Months to years

About

Background: Physics, software automation, and machine learning applied to experimental science.

What I Do: I design and build closed-loop experimentation systems – combining real-time sensing, automated control, and ML optimization to explore chemical parameter spaces faster than traditional methods.

Core Capabilities:

  • Transform traditional analytical techniques into ML-optimized systems
  • Spectroscopy (UV/Vis, Raman, IR, fluorescence, NMR)
  • Mass spectrometry and flow cytometry automation
  • Imaging systems (hyperspectral cameras, computer vision, microscopy)
  • LiDAR and 3D sensing integration
  • Real-time monitoring with edge computing pipelines
  • Microfluidic/flow reactor integration and control
  • ML-guided experimental design (Bayesian optimization, active learning)
  • ML-powered nanoparticle quality control (real-time size/shape/quality prediction)
  • Nanoparticle synthesis optimization and process control
  • Legacy instrument modernization (sensor retrofitting, cloud connectivity)
  • Data pipeline architecture for experimental workflows

My Approach: Craft Shifted by Automation

I believe in high-throughput experiments that are lightweight. FTEs should go from manual experiment to scaling experiments.

ML and AI can fill in human gaps with unbelievable precision. That doesn't mean craft disappears—it shifts. Some parts of chemistry will be done by AI. The new craft is guiding these experiments, and very importantly: building the hardware to make it scale.

I'm not interested in building models for general chemistry. I want to solve very specific problems chosen intentionally. It's signal vs noise.

What Sets This Work Apart

I bring automation, ML, and systems engineering expertise directly to chemistry challenges. This combination lets me rapidly build measurement and control solutions that unlock new possibilities—transforming "we can't measure that fast enough" into "now we can."

The impact: Most chemistry innovation is blocked by measurement and control limitations. I specialize in removing these barriers.

Let's Talk

Whether you need automation expertise for your R&D, are interested in autonomous chemistry research, or exploring deep tech in chemistry automation.

For Consulting Clients

Need automation expertise for your R&D?

Schedule 30-min consultation

For Collaborators

Interested in autonomous chemistry research?

Let's talk

For Investors/Advisors

Exploring deep tech in chemistry automation?

Get in touch

Email: sjoerd@datadrivenlabconsulting.com
LinkedIn: linkedin.com/in/sjoerddehaan