In 2007 I started work at Stainless Games located on the Isle of Wight. You might have heard of them as the creators of the Carmageddon series and the project is very much cherished here. As I write this I can see a desk full of awards for it and several boxes of the game standing not 20 feet away from me.
The first Project I worked on was an unannounced title for a publisher which unexpectedly hit some restructuring problems and the project had been dropped before it went into full production. For obvious reasons (at least I hope they're obvious to you, because I'm not sure about them myself) I can't really talk about details and have to be really vague about it. It was a remake of a 20 years old strategy game and I was making some initial models fot it and some general mock-up screens.
After not long I was moved to work on Red Baron Arcade. A project destined to suffer from some publisher issues as well but more about that later.
When I got on board the project was already in full production and the game was perfectly playable. After finding my place in the project and doing all sorts of odd jobs my first proper modelling job on it was making sea units. 
Despite the Arcade nature of the game, the planes and other units were kept realistic and were based on old photos and blueprints of the era. Certain liberties were taken to make some easily distinguishable but if you look at the accompanying photos you'll see they're pretty closely adhered to.
My first model was the British destroyer. I didn't have the blueprints for it so I was working from hundreds of photos I had. Unexpected help came from the many miniature models companies make of ships from all periods. Miniature models are usually cleaned up from all the minute deck equipment cluttering up the decks and make it easier to get the underlying shape.
Obviously the ships weren't the main 'characters' in the game - it being a game about planes (and bi-planes and tri-planes) - so they obviously had a limited budget of polys and textures. Each usually used two small textures and had the light map baked into the diffuse texture. The smaller texture was used at lwoer density on the underbelly which was obscured by water most of the time. The ships also required a compliment of particle effects created in the in-house particle tool which made creating smokestacks and explosions fairly easy and I'd be coming back to use it quite a few times during the project to create exploding buildings and smouldering wreckages.
As for the tools used, having moved to Stainless I switched back to 3D Studio Max as that was the package most people on the project used. Photoshop was used for both diffuse textures and normal maps. I remember being an annoying zealot and 3D Studio Max advocate in my younger years but by the time I got to Stainless I've become pretty much software agnostic and comfortable in whatever package the company prefered or had the license for.
It happened that most of my work was concentrated on the map featuring a coastline. This included the manual labour involved in creating a mesh for the coastal surf effect spanning the huge map. However much automation you put into it you always end up having to tie it all up by hand in the end when the coastline takes sharp turns or has tiny nooks and crannies you have to compensate for.
Part of my the coastal renovation effort was making wooden piers and and jetties protruding from the coastlines and accompanying cranes and containers (which I learned haven't been standardised until the 1950s - we spared no effort on our research) and building sets of warehouses and even dry docks you get to blow up in the game
I also got to work on other maps adding detail to some missions by scripting in vehicles driving around the maps and adding aforementioned smouldering wreckages and burning buildings. All this was part of a collaborative effort and not possible to isolate so there's little point in showing the individual models.
Work on the maps was done in an in-house editor - most of my work was done on the coastal map. You can't really call it level design as the ground was in the game mostly a backdrop for the action which took places in the skies. I only mention it because it's the first time I got to use Perl in gamedev, even if it was in a strictly auxiliary capability.
Games development is 99% of the time a strictly windows based business - even if you make games for consoles. Perl hasn't really got a place in this world. Scripting languages are used, but the the only one receiving widespread attention and use is Lua. Otherwise companies tend to use their own internal scripting languages, like Stainless did before adopting Lua itself. The Perl script was not in any way used by the game or the editor - it was a just a helper script for me and it's simple job was to remove overlapping duplicate objects from maps by munging the level file (and there's nothing better at munging text files than Perl).
Not very exciting but the OpenPerl IDE (a long-time defunct windows IDE I used since uni which I kept using until around 2009 when I finally switched to Strawberry Perl and Padre for working in Windows) stayed with me to help me in such tasks ever since - especially when I got to work on the front end which was controlled by giant cyborg-readable XML files.
Taking over the work on the front end was to be a sign of things to come. Red Baron's front end was heavily reliant on the C side and was little more than a set of XML files being parsed by the game engine and any interactive capability was coded in C so I didn't get to do any actual coding yet. My work was to control it through the XML files and provide the assets for the buttons, frames and backgrounds used throughout. So maybe it was less than a glorious experience but by golly was it character building and vocabulary enhancing (though mostly by expletives). It was a terrific primer for working with front-end systems though and as crash courses go this was the meteor that wiped out the dinosaurs.
From then on I'd be working less and less on the graphics side and concentrating on the front-ends and other code-art intersections as the company scrapped the XML controlled system for a sleeker and more powerful Lua based one. I'd get to see it born and grow up into a substantial part of the company's tech despite (or, as I like to think, thanks in part due to) my constant abuse of it, to the bemusement and almost paternal worry of its author - a fine gentlemen who I hope doesn't hold a grudge against me for it.
We finished the project, got it through cert and then the worse that could happen happened. The publisher, Sierra Online, went under and during the turmoil the project just sat there waiting for someone to turn the greenlight on for it. 
After such a long time in limbo when we thought it was gone for good Activision who bought it out finally signed it off and it got released but only on the PS3 and the PC. The economical crisis was in full swing and worse things were to come. 