jlm·technologies

/ JLM Technologies · Cape Town, South Africa · 33.92°S 18.42°E

Joshua
Malan.

Mechanical & Biomedical Engineer turned one-man data platform. I build financial engines, devise fleet telemetry, Machine Learning valuation models and privacy enforced database systems. Every number on this site was proven twice.

press ~ anywhere
for the terminal

100k+

lines of code shipped solo

0.90

area under the curve, payment prediction (1.0 = perfect)

99%

revenue reconciled to billing

5 000+

access points under telemetry

2.14M

transactions analysed

0

residual radio channel conflicts

Selected systems

01in production · internal

Financial Engine

A bottom-up cash-flow model that reconciles to 99% of billed revenue.

99% revenue reconciliation0.90 AUC, payment prediction (1.0 = perfect)73 golden parity tests
open
02codename · confidential

VEIL

Privacy enforced by the database, not the application.

2,869 lines of SQL enforcement19 persona-matrix test files3 disclosure tiers
open
03pilot

Valuation Engine

A machine-learning pricing engine that reports the worse but honest number.

99.6k market listings modeled16.5% average pricing error, honestly measured (was 18.4%)80% price-range coverage, as designed
open
04in production · internal

RF Telemetry Platform

10,000+ access points. Zero residual channel conflicts.

10,000+ access points, 8 regions0 residual same-channel conflicts25k lines of analysis code
open
05in production · internal

Telemetry Ingest Pipeline

A database outage loses nothing. Proven rather than promised.

10,000 radios capacity-planned80 tests, zero-loss proven3 duplicate-prevention layers
open
06live

Hiring Platform

Apply for a job from any phone. No account, no CV, four languages.

4 languages, enforced at build time3 apply paths, AI-optional5 days to first live pilot
open
07live

Price Comparison Engine

Photograph a competitor's invoice. Get an honest savings report.

3 matching tiers, each match traceable2 VAT bases, computed fairly40 tests on the pure core
open
08msc research · uct

Myocardial Mechanics

Measuring how heart muscle deforms, and making the material model prove it.

100+ tissue samples tested80+ imaging datasets analysed10x10 mm specimens
open

joshua@jlm-technologies:~$ ls projects/

How I work

Five rules that show up in every project, usually because breaking them once was expensive.

P1

Honest metrics or nothing

When a champion model's 10.6% error turned out to be a measurement flaw, it was rebuilt on a protocol that measured 12.3%, and had to win on that number before shipping. Experiments that prove nothing get reported as proving nothing. The customer-facing report shows where the competitor wins.

P2

Parity as proof

Every port from one language to another is verified against golden fixtures from the original: to a millionth, to the cent, or bit for bit. One implementation per metric. A rewrite that cannot prove it behaves identically does not ship.

P3

Enforce at the lowest layer

Privacy in database policies rather than application code. Second-factor checks inside the identity resolver. Builds that fail when a language is missing a translation, or when a development shortcut reaches production config.

P4

Fail closed, degrade gracefully

Rows are absent, never redacted-but-present. Ingest buffers through database outages without loss. When AI budgets run out, deterministic paths keep the product working.

P5

Audit yourself adversarially

Commissioned audits of my own work found 20 wrong-number discrepancies in a dashboard port, a critical authorization bypass, and two data-mixing bugs in machine-learning models. All published internally. All fixed with regression tests.

The arc

I trained as a mechanical engineer: biaxial testing of heart tissue, digital image correlation, a finite-element solver written from scratch in Python. The rigor stuck.

Then I joined a township internet provider as a one-person data function and taught myself the modern stack on the job: data engineering, statistics, machine learning, full-stack product, and the operations to run it all in production. Everything on this site was designed, built, deployed and documented by me, then reviewed and audited adversarially, with parity tests as the referee.

Alongside that, I ship real products: an AI hiring platform live in pilot, a machine-learning valuation engine in dealer pilot, and a privacy architecture I can only show under a codename.

/ what I live by

Stoicism, practised rather than quoted: control what I can, meet the rest with composure. Hard work as the baseline, never the brag. And courage: publish the honest number, start the hard thing, be a beginner in public.

/ experience

Nov 2025 – present

Data, ML & Platform Engineer · Ikeja

The entire data platform, analytics, machine-learning and internal-tools function for a South African township internet provider, as one person. Built the bottom-up financial engine, the radio telemetry and optimisation platform across a fleet of more than 10,000 access points, the push-telemetry ingest pipeline, and the full-stack analytics platform that replaced the previous dashboard stack. Schema to ingestion to machine learning to product to deployment.

2024

Postgraduate Tutor · University of Cape Town

Tutored Solid Mechanics, Mechanical Vibrations and Machine Elements Design.

2022

Engineering Intern · Jonkers Sailplanes

Operational analysis across production and administration. Design and fabrication of fibreglass and composite glider components.

2021

Engineering Intern · Potchefstroom Engineering

Concept-to-execution mechanical projects: CAD to CAM to CNC manufacturing and shop-floor implementation.

/ education

MSc

2023–2025

Mechanical & Biomedical Engineering, University of Cape Town

Completed. Myocardial tissue characterisation: biaxial testing with full-field optical strain measurement, custom 3D-printed apparatus, and Python pipelines for stress computation and material-model fitting. Also built a from-scratch Python finite-element solver covering elasticity and heat conduction, linear and nonlinear.

BEng

2019–2022

Mechanical Engineering, University of Pretoria

Dean's Merit List 2020. Tutored Solid Mechanics, Vibrations and Machine Design.

NSC

2014–2018

Pretoria Boys High School

Five distinctions.

ECSA

Candidate Engineer

Registered with the Engineering Council of South Africa.

Off duty

try `personal` in the terminal

Joshua Malan in a tuxedo on his wedding day
/ the wedding
Katiya under her veil on the wedding day
/ the better half
Joshua Malan in a gi after his first BJJ tournament
/ first tournament
music

Classically trained pianist, ages five to eighteen, theory and all. A passion for music from as early as I can remember.

reading

A reader first, classics mostly. Currently deep in Dostoevsky and Camus, with Nietzsche and Kafka in rotation and the Stoics on the nightstand.

bjj

Brazilian jiu-jitsu blue belt. Sounds impressive until you watch me get folded daily by smaller, calmer people. I keep going back, which is the whole point.

languages

English and Afrikaans fluently, and enough Portuguese to follow an argument about football.

linux

Pop!_OS daily, Arch for tinkering. The dotfiles folder has commit history.

football

Manchester United. The accent colour on this site is not a coincidence.

family

Married to Katiya Carreira. Family first, everything else after.

References

contact details on request

Peter Coetzee

Owner, Ikeja. Current employer.

Dr Reuben Govender

Postgraduate supervisor, University of Cape Town.

Dr Niel van der Watt

Character reference.

/ contact

joshualiammalan@gmail.com

Open to hard problems, especially where data, safety-critical engineering and honest numbers meet.