Work

Systems shipped under real constraints.

W.01 · ML detection pipeline

Real-time drone detection from radar streams

Detect and classify drones from raw radar tracks in milliseconds, at a false-positive rate low enough to be operationally usable.

97.2%
recall: 100%
matrix test detection: -84% (V1 to V3.2)
false positive reduction: 0 after ONNX
runtime segfaults: 6
classes

W.02 · Solo, mobile to data pipeline

Correct, cheap brand attribution from one raw stream

Attribute every social post to the right brand, accurately and cheaply, from a single unfiltered stream.

99%
attribution accuracy: 3,725
posts per pull: ~16.7M
total reach: ~600 green
backend tests: 97%
live probe attribution: live on TestFlight
status

W.03 · Full-stack, from scratch

Canes: a from-scratch e-commerce platform running in LIVE mode

Ship payment, shipping, stock, a multilingual storefront and an admin panel as one self-sufficient platform, with no marketplace lock-in.

~48
data models: 2 (customer + admin)
panels: 4
live integrations: LIVE (real payment + shipping labels)
mode: 28MB to 1.9MB
homepage image weight: 64 to 85+
mobile speed score

W.04 · Library design and implementation

KickForge: two streaming protocols behind one pip install

Turn the OAuth, WebSocket and webhook plumbing for stream bots into one install-and-auth toolkit, with no public URL.

~2 min, 3 steps (install, auth, run)
time to bot: 5 modular (core, bot, clip, gsi, overlay)
packages: one PyPI package
distribution: v0.3.0 live
version: none
public url needed

● available for contract work

email →linkedin →