Bohea Still · independent industrial software engineer · Shenzhen, China
Pick your question
2.5 years as sole developer of every internal system at a robot maker. Signed CTO reference, on request.
Operator HMIs, undocumented devices, machine data into your customer's ERP.
7 case studies and 19 public repos that run on your laptop in a second.
Fixed scope, milestones, proper invoices, and you own the source outright.
Four kinds of work I turn down, and one kind of buyer I'm wrong for.
≈230×
less data scanned: 1.9 GiB → 8 MiB, same query, measured by query_log
50 → 5,000+
QPS after rebuilding a panel maker's ERP/WMS, downtime down ~90%
100%
of wiped production data recovered, by reverse-engineering MySQL binlogs
I work 10:00–20:00 UTC — the whole European working day and the entire US morning. A reply within 12 hours, always.
Bohea Still
Independent industrial software engineer · Shenzhen, China · [email protected]
The night shift isn't an accommodation — it's when I do my best work, and it happens to land on your working day. I work in writing on purpose: precise, searchable, and it survives a timezone. Calls when a decision is genuinely stuck.
HKC Corporation — display-panel manufacturer · HQ IT, building the company's internal ERP and WMS systems.
I re-architected a fragile monolithic ERP/WMS into event-driven microservices, sitting with the business teams from spec to launch — downtime failures down ~90%, throughput 50 → 5,000+ QPS. Talking every day to the people who actually used the system is how I learned to build what a business needs, not just what a ticket says.
Shenzhen Turing Robot · document understanding for banking and securities clients, years before ChatGPT. One of the engineers on it.
Extracting tables out of PDFs — borderless tables, one cell spanning lines, several cells sharing one. Open-source tools got about 55% of them right. We fused two engines, matching each cell to its characters' coordinates by text similarity, and reached 99.5%; adding a document-conversion path took multi-format fidelity to 99.99%. Not comparable to today's LLMs, but the lesson holds: engineering rigour plus domain understanding beats a raw model. (Still on good terms with its founders.)
Shenzhen Coming Financial · a paid consumer platform for investor and roadshow events. One of 8 backend engineers.
I led production database migrations across seven microservices and owned high-concurrency paid consumer flows. Delivering large, non-standard work reliably — in phases, milestone by milestone, without dropping the ball — is the discipline I built here.
Standard Robots, Shenzhen — an industrial AMR maker · sole developer for every internal system the company ran.
I built and held everything the company ran on except the robot: BOM and supply chain, CRM, expenses, file storage, deployment. Along the way I saw how a configuration becomes a bill of materials, how one model variant multiplies into part numbers, and where after-sales calls really come from. I also recovered 100% of deleted production data by reverse-engineering MySQL binlogs, with operations still running. Backed by a signed letter from the CTO, available on request.
Betting my craft on my own name. All in — this is what I do now.
My permanent home base — every project and relationship centers here, not on any single platform.
Integrators and machine builders win the machine and run short on software people. I cover everything between "the machine arrived" and "your customer's people run it every day". The operator interface is just the visible part — and the three layers below bite in this order.
One-page PDF if you're passing me on to someone.
You ship 20 machines a year and maintain 20 interfaces. Each model has its own screens, its own quirks, its own commissioning days on site, and every support call starts with "which version is it?" I build one operator interface that configures per model, with PLC, Modbus, OPC UA or MQTT underneath, plus the device integration and the remote support path. Commissioning shrinks and your team stops maintaining variants.
I also speak ERP, because I spent two years rebuilding one (BOM, work orders, batches, traceability) before I ever touched machine software. The trade calls this layer shop floor to top floor; in ISA-95 terms, Level 2 to Level 4. I do that work. I don't sell you an MES. So when your customer asks whether the machine's data can land in their system, that's a yes you can sell, not a second vendor you have to go find.
A client actually asked me: "Our field engineer adjusted the tool parameters and renamed the points himself, so what exactly did your work buy us?" My answer → · If you're an integrator or an OEM, there's a page written for you →
"It runs, but the operator can't run it." One operator interface configured per model instead of a screen-set per machine; devices that were never meant to be read, read anyway: Modbus, OPC UA, MQTT, serial, vendor SDKs; alarm and interlock paths that fail loudly instead of quietly.
Live now: an operator HMI for an industrial robot, for a North American integrator. And a public repo where 11 attempts to make a machine hurt someone all get refused, in about a second.
"The shop-floor count doesn't match the ERP count." Work orders going down, WIP and batches coming back up, traceability, reconciliation. ERP people don't touch machines; machine people don't know what a work order means to the business. Nobody owns this layer. I've been paid on both sides of it.
93% of manufacturers have an MES; only 23% have actually integrated it with ERP, quality and OT systems. The gap is that layer. (Rockwell Automation survey, Jul 2026, 1,560 decision-makers.)
"You're not on site — so how do I know it works?" I build against a simulated device first, so the software is finished and tested before the real machine exists. Mock-mode demos prove the whole path before I hold a single credential. Improvements get pinned in CI so they can't quietly regress, and every milestone ships with written, screenshot-backed verification.
The front half of this has an industry name: virtual commissioning. Fixing a machine issue during commissioning can cost up to 100× what it costs in design (Rockwell Automation), so I quote that front-loading as its own line item. See it done: the ClickHouse case below is measured by the database's own query_log and guarded by a CI gate, and the demo library runs credential-free.
If none of these is your sentence, say yours, including "an AI wrote this and now nobody can touch it", which I hear a lot and do take. I'll tell you honestly whether it's mine to solve, and if it isn't I'd rather say so than take the work. If the question is whether it's possible at all, there's a three-day sprint for that →
These are tools, not the pitch: PLC · Modbus · OPC UA · MQTT / Sparkplug B · operator HMI stacks · ClickHouse · MySQL binlog · event-driven services.
Every project below is backed by something you can verify: a running demo, measured numbers, or handover docs. The first one is where I spent two and a half years. Work I did for other kinds of buyers is folded at the bottom: real, just a different lane.
Work I did for other kinds of buyers — data extraction, payments, marketing ops, routing — lives in the archive. Real work, different lane.
Beyond these deep dives: a shelf of runnable public demos — credential-free, each honest about what it proves. Browse the demo library →
Not "is he good" — that part you can check yourself. The real worry is: he's unreachable in my hours, my finance department can't pay him, and if he vanishes I'm stranded. Answers to all three, plus the two that follow from them.
You see working software before you spend a dollar on implementation. Mock modes let me prove the mechanics without touching your credentials. And where automation truly needs a login, it still doesn't need your password: a session you sit next to, or a throwaway credential you can revoke.
No radio silence. I work async and send short Loom updates showing exactly what's built, with bugs and risks surfaced early, not hidden. Docs ship with each milestone too, because documentation written at the end doesn't exist on the day someone goes quiet.
I've reverse-engineered MySQL Binlogs to recover 100% of lost production data during an emergency. When something goes wrong, I'm the person who stays and fixes it.
Software I deliver does not implement your safety interlocks. E-stop, guard doors and light curtains belong in a safety relay or a rated safety controller — hardware that keeps working when my code is wrong. I build the layer above it, gate it hard, and say so in writing. An engineer who offers to put your E-stop in the application layer is telling you something about their experience.
Source can be committed to your repository from day one, not packaged at handover: stop at any point and you're holding something complete and runnable, so your exposure is never more than one milestone. Handover docs cover how to run it, how to migrate it to your own accounts, and every credential to replace. No lock-in, no dependency on me.
And what I don't do
What working with me actually looks like — the terms, the paperwork, and how to start.
The terms
How I quote
Fixed scope, tiered quotes, milestone payments. The price I quote is the price. If your budget is tight, say so in your first message: I'll tell you straight away whether it fits, instead of letting you find out three calls later.
How my current engagement started, for reference: a first project at $3,000, delivered; a second at $10,000, now in progress, same client. The second decision is the signal.
Payments & paperwork
Together they're everything your accountant needs to reconcile the payment without friction — without chasing anyone. Expand only what you want to see.
Scope, milestones and deliverables in writing, signed by both sides before any work starts.
A clean PDF invoice per milestone — your company name, the amount and the project, ready to file.
Payment runs through Payoneer (corporate card or bank transfer); you receive Payoneer's own official record confirming it.
US clients: I provide a W-8BEN, so your tax and reporting side is handled. One person to deal with — paperwork that runs like a proper vendor's.
If your procurement or compliance requires contracting with a corporate entity, I can arrange that before we sign.
First time buying software this way? Most clients book it under project implementation or fold it into commissioning; ongoing work goes in the maintenance line. Happy to write the invoice so it lands where it needs to.
Passing me along to someone?
Say this: "I know someone who does the software that ships with machines — operator screens through to machine data landing in the customer's ERP. Remote, and the code goes in your own repo." Then send them boheastill.com, or forward this one-page PDF without writing anything: the line to describe me with, three things they can verify, and how to reach me. You shouldn't have to summarise me — that's my job, not yours.
Not sure it's even possible?
If the question is "can this be done at all", there's a scoped way to find out without committing to a project: a three-day Feasibility Sprint — a written answer, a running demo if it's feasible, half back if it isn't. If you already know what you want built, skip it and just ask me to quote the scope; that's free.
Book 20 minutes
Pick a slot. You don't have to write an email first explaining the situation. Before we hang up you'll have three things: your problem played back to you (so you know I heard it), my read on which layer it's actually stuck in, and a rough price tier. No deck, no "discovery phase", no follow-up sequence. If it's not a fit, I'll say so on the call.
The calendar shows times in your timezone automatically, and the invite arrives with a Google Meet link. Slots cover the whole European working day and the US morning.
Start here
Tell me what your platform can't do, and what it's costing you. I'll reply with an honest read on whether it's solvable — and if it is, a demo usually follows.
Don't feel like writing an email? Leave your problem and any way to reach you (email is best) — I'll get back to you, usually within a day.