About

Surapat (Arm) Ek-In, PhD

From particle physics to production systems. I did my PhD hunting for new physics at the Large Hadron Collider. I’ve spent 8+ years across scientific data pipelines and production software — including full-stack engineering where my software shipped to more than ten customer sites. Now I apply the same method to your systems.

Experience

  1. 2026 — Present

    Independent Consultant · ArmLab

    • Altruistic AI: agentic LLM pipeline (LangChain, RAG, DuckDB, Azure) — fewer hallucinations, faster time to first token.
    • Engram.health: multi-source data ingestion and normalisation pipeline.
    • Python
    • LangChain
    • RAG
    • DuckDB
    • Azure
    • FastAPI

  2. Feb 2023 — 2026

    Software R&D Engineer (Full-Stack) · Lino Biotech

    acquired by Miltenyi Biotec

    • Core software for the MACS Matchmaker biosensor, shipped to 10+ customer sites.
    • 10× faster real-time analysis through profiling and async/concurrent code.
    • 5× sensor sensitivity by redesigning the analysis and imaging pipeline.
    • Owned the full stack: Python/FastAPI, React, hardware control, Jetson Orin edge.
    • Python
    • FastAPI
    • TypeScript
    • React
    • NVIDIA Jetson

  3. 2018 — Jan 2023

    Experimental Particle Physicist · CERN — LHCb / EPFL

    • Led the data pipeline for the first observation of the charm-meson mass difference (PRL 2021).
    • Model-independent charm-mixing measurement (PRD 2023); 4× lower systematic uncertainty.
    • Signal extraction over ~1B-event datasets on the computing grid.
    • Low-latency C++ detector tracking with neural-network components.
    • Python
    • C++
    • ROOT
    • RooFit
    • SLURM
    • Grid Computing

  4. 2021 — 2022

    Lead Data Engineer / Data Scientist · Altruistic Innovation Ltd

    part-time · energy / smart-grid

    • Micatu (smart-grid optical sensing, USA): sensor model estimation and correction.
    • Turned client constraints into fast prototypes for production ML.
    • Cut temperature-induced noise in optical edge sensors for smart-grid monitoring.
    • Built ML pipelines and AWS cloud architecture (SageMaker, S3, EC2).
    • Python
    • AWS
    • SageMaker
    • ML Pipelines

  5. 2015 — 2016

    Undergraduate Research Student · Mahidol University

    Department of Physics, Bangkok

    • First publication: magnetic reconnection in astrophysical plasmas (ApJ 2017).
    • Astrophysics
    • Plasma Physics
    • Numerical Simulation
Download full CV (PDF)

How I Work

Measure first

I profile before I optimise. At CERN I ran signal extraction over ~1B-event datasets. At Lino I profiled the real-time analysis loop and cut it 10× with async and concurrent code.

Full stack

I build whole systems end to end: typed Python and FastAPI backends, React and Next.js frontends, hardware control, data pipelines, and edge inference on NVIDIA Jetson Orin.

Research → production

I've published in Physical Review Letters and Physical Review D. I've also taken lab prototypes to market: Lino's MACS Matchmaker biosensor shipped to paying customers across 10+ sites.

Independent partner

I work part-time and embedded, usually next to your own engineers. I'm one person with no agency hours to fill, so I'll tell you when a fix is simple or when you don't need me at all.

"Most of my work at CERN was isolating one signal from a billion noisy events. Production systems are friendlier than that."

Contact

Let's build together

I help build and improve production systems, monitoring and traceability, and hardware-software reliability.

Start a Conversation