CRAWL_INFRASTRUCTURE · IN_PRODUCTION
APEX.
A crawl platform for targets engineered to refuse extraction. Not a collection of scripts — a layered system where fingerprinting, identity, solving, and egress are separate, tested components. Data products run on top of it; custom pipelines for clients are built from the same layers.
Architecture
Four layers.
Anti-bot toolkit
TLS fingerprint shaping (JA3/JA4) via BoringSSL-backed clients, header-order discipline, HTTP/2 settings matching. Requests leave the platform indistinguishable from the browser they claim to be.
Identity & session layer
Persistent personas with stable session IDs, trust aging, and cookie jar continuity. Sessions accumulate reputation instead of burning a fresh identity per request.
Browser-solver bank
A Python farm that solves challenges once and banks the tokens — CAPTCHA clearances and signed tokens are reused across the fleet instead of solved per request.
Residential proxy mesh
Self-managed exit nodes with reverse-tunnel enrollment. No third-party proxy dependency, rotation policy owned end to end.
Open_Source · Components
The public layer.
Detection, transport, and mobile-RE components of this platform, published and pinned on GitHub. The mesh, escalation policy, and orchestration stay private.
Classify which anti-bot vendor protects a target — and which layer actually blocks you.
Check any client's TLS fingerprint (JA3/JA4) against real browsers and flag incoherence.
Production-hardened curl-impersonate wrapper for Node: discovery, coherence guard, doctor.
Solve a challenge once, replay the clearance a thousand times. Browser farm + warm token bank.
APK in, API spec out. Static extraction plus Frida/mitmproxy dynamic capture.
CI tripwire that alerts when a target's defenses change — before the pipeline breaks.
In_Production
Running on it.
Data Vertical · Food Delivery
Restaurant & venue intelligence.
The flagship pipeline starts where the public web ends: the data lives in a mobile app, not HTML. The API was reverse-engineered from the APK — JADX for decompilation, Frida and objection for SSL pinning bypass — then rebuilt as a production crawler on the APEX layers above.
Thousands of venues tracked daily across major metros, with every menu, price, and availability change recorded as a queryable event stream.
Pipeline_Output
Data Vertical · Automotive Classifieds
Vehicle market intelligence.
Recurring crawls across Europe's major car marketplaces — targets behind Cloudflare, DataDome, and Fastly Bot Management. Several of these stacks had stopped the internal crawling teams of a European automotive data group, including one target where a lead engineer's three-month solution never stabilized. APEX serves those edge cases in production.
The same four layers, pointed at classifieds: fingerprint-shaped transport, identity continuity, and escalation only when the target demands it.
Pipeline_Output
CUSTOM_PIPELINES
Need data from a target that resists it?
The same four layers get pointed at your target — site or mobile app, behind Cloudflare, DataDome, or PerimeterX. Fixed scope, senior-level only.