We're the system of record for the Canadian vehicle: raw automotive data turned into facts that can be checked. Our open-source VIN decoder sits underneath a government vehicle registry.
Some of what's interesting:
- A government registry, a manufacturer's own filing, a listing site and an insurance record will describe the same truck four different ways, and nothing in any of those files says they're the same truck. Reconciling that, for every vehicle in the country, in a way that holds up when a regulator asks how you know, is most of the data work.
- Seventeen characters on a door frame encode manufacturer, plant, model year and build attributes — badly, inconsistently, and differently per manufacturer. Public reference data covers some of it. The rest gets worked out from filings, recall documents, and patterns inferred across millions of vehicles.
- We read vehicles directly over OBD-II now. Forty modules answered on a recent Hyundai, each reporting its own part number and installed software version. Nothing downstream of the VIN moves when a module is reflashed, so no registration, history report or dealer system knows what software is actually in a car.
Cardog | Toronto, ON | Remote/Hybrid (Canada) | Full-time
AI-powered automotive platform — we’re building the infrastructure layer for vehicle data, from VIN decoding to market intelligence, powering government and public-sector vehicle programs.
Looking for: Automotive experience, Full-Stack, AI/ML, Data Engineers (public-sector/govtech experience a plus)
Cardog | Toronto, ON | Remote (Canada) | Full-time
AI-powered automotive platform — we're building the infrastructure layer for vehicle data, from VIN decoding to market intelligence.
Looking for: Automotive experience, Full-Stack, AI/ML, Data Engineers
not sure what PocketOS does or why your whole dataset would be a single volume without a clear separation between application and automotive data. how are you decoding VINs?
Makes me wonder also about multi-tenancy. If all customer information is in single volume. How big risk they put on their customers on their most business critical and proprietary data to leak other competitiors?
Cardog | Toronto | Full Stack Engineer | In-person
We build automotive software. We work with dealerships on ad-hoc projects, websites, and pricing tools. We also run consumer facing products - a web app for vehicle shopping, research and market data, and a mobile app.
Right now we're building EV battery health diagnostics - connecting to a car's BMS over CAN bus, pulling cell voltages and state of health, generating reports. Mix of hardware and software.
Small company with real clients and real revenue. One engineer. Looking for a second who wants to own things e2e.
The need for proxies in any legitimate context became obsolete with starlink being so widespread. Throw up a few terminals and you have about 500-2k cgnat IP addresses to do whatever you like.
The actual secret is to use IPv6 with varied source IPs in the same subnet, you get an insane number of IPs and 90% of anti-scraping software is not specialized enough to realize that any IP in a /64 is the same as a single IP in a /32 in IPv4.
> any IP in a /64 is the same as a single IP in a /32 in IPv4
This is very commonly true but sadly not 100%. I am suffering from a shared /64 on which a VPS is, and where other folks have sent out spam - so no more SMTP for me.
If they're CGNAT then unless Starlink actively provides assistance to block them it won't matter.
As someone who wants the internet to maintain as much anarchy as possible I think it would be nice to see a large ISP that actively rotated its customer IPv6 assignments on a tight schedule.
why do we insist on bringing the worst tech from sci-fi to reality? is our collective intelligence still not enough to thwart the social and market dynamics?
Come on, we are a race that took no less than 2,000 years (!) to realize the seat of thought lies in the brain and not in the heart, counting from when a prominent figure first hypothesized it (Alcmaeon of Croton).
all of these problems are better articulated at the level you just explained them. the code for these issues is convoluted and is only of use when an entity (human or not) can actually manipulate the symbolic text that achieves that task. a random oauth stub is of 0 use to the most skilled programmers without documentation as to what contracts and invariants are. bits in a file is just a means
We're the system of record for the Canadian vehicle: raw automotive data turned into facts that can be checked. Our open-source VIN decoder sits underneath a government vehicle registry.
Some of what's interesting:
- A government registry, a manufacturer's own filing, a listing site and an insurance record will describe the same truck four different ways, and nothing in any of those files says they're the same truck. Reconciling that, for every vehicle in the country, in a way that holds up when a regulator asks how you know, is most of the data work.
- Seventeen characters on a door frame encode manufacturer, plant, model year and build attributes — badly, inconsistently, and differently per manufacturer. Public reference data covers some of it. The rest gets worked out from filings, recall documents, and patterns inferred across millions of vehicles.
- We read vehicles directly over OBD-II now. Forty modules answered on a recent Hyundai, each reporting its own part number and installed software version. Nothing downstream of the VIN moves when a module is reflashed, so no registration, history report or dealer system knows what software is actually in a car.
Hiring: VIN Decode Engineer, Data Platform Engineer, Infrastructure Engineer, Founding Engineer (dealer platform). TypeScript, Postgres, Cloudflare Workers.
Full descriptions and apply: https://cardog.app/careers
reply