Germanium photoconductors offer excellent sensitivity in the 50-140 m spectral range. Coupled with their modest cooling requirements and their compatibility with the silicon cryo-CMOS readout technology, these detectors are the most attractive candidates for far IR astronomy in this wavelength range. Over the years we have been pursuing the advancement of this technology and our initial effort has produced a 2x16 Ge:Sb array with an NEP in the low 10 -18 W/ Hz range, rivaling the best far IR arrays currently available. Further work has resulted in design and fabrication of a low noise, 2-side buttable 32x32 (64x64 mosaic) CTIA readout, the first 1k-pixel Ge:Sb fully assembled focal-plane array, a new hybrid design better suited for far IR photoconductors, and the preliminary design of a 2-side buttable 64x64 (128x128 mosaic) CTIA readout. Our developmental work continues and we believe that sensitivity levels below 10 -18 W/ Hz are within reach. This paper presents an overview of our progress so far and outlines our roadmap for further work.The development of far IR focal-plane arrays (FPAs) is at the heart of technology needs for NASA's future astronomical missions. In the 50-140 m wavelength range, germanium FPAs have been the arrays of choice due to their low noise, modest cooling requirements, and compatibility with the established silicon readout technology. Their success is demonstrated most recently by instruments such as Spitzer's MIPS [1,2], Herschel's PACS [3,4], and AKARI's FIS [5,6]. To further this technology, we have been making incremental progress in areas of material processing, cryogenic readouts, and integration of the readout and the detector with the ultimate goal of developing the capability similar to that which currently exists for shorter wavelength IR arrays. The path to establishing such capability is not straightforward since practically all the principal components of such an FPA need to be developed and, due to significant differences between short and long wavelength photodetector materials, the technology for short wavelength FPAs cannot be extended routinely.As the first step, we developed a prototype 2x16 FPA using Ge:Sb detector and a 1x32 (SBRC190) CTIA cryogenic readout [7]. By achieving a noise-equivalent-power (NEP) of 2.2x10 -18 W/ Hz, this FPA demonstrated the potential of this technology for future far IR instruments. However, in order to take full advantage of large apertures of future space and airborne telescopes such as SOFIA, SAFIR/CALISTO [8,9], and SPICA [10], FPAs at least as large as 10k-pixels are needed. To produce large format FPAs reliably, two key technologies need to be developed: a) a low noise, two dimensional CTIA readout and b) a planar, bump-bond hybridization process suitable for far IR photoconductors. We started by design and fabrication of SB349, a 32x32, 2-side buttable (64x64 mosaic) CTIA (capacitive transimpedance amplifier) readout multiplexer [11], by far the largest of its kind. The latest test results, a summary of which is presented in this...