1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
|
\subtitle{Projects}{\faSuitcase}
% While many projects are spread over time and going in parallel, I have tried to specify shorter core development times.
\begin{projects}
\ifdef{\scivariant}{
\project{Since 2018} {KaaS} {KATRIN-as-a-Service (KaaS) platform for distributed data acquisition systems}
\action{\since{2018}} {4} {Prototyped platform, secured funding, built team, and led rollout}
\action{\since{2018}} {4} {KaaS cloud is integrating data from fast detectors and thousands monitoring sensors}
\action{\ivl{2018}{2019}} {} {Cut costs by adapting legacy software for distributed ops and integrating HPC resources}
\action{\since{2022}} {1} {Led development of AF\_XDP-based, containerized DAQ framework}
\action{\since{2020}} {3} {Modeled data access patterns and guided research on organization of time-series data}
\action{\since{2015}} {1} {Supervised development of online monitoring framework for the KaaS ecosystem}
% \action{\ivl{2019}{2020}} {3} {Coordinated migration of the KATRIN services to the KaaS platform}
% \action{\since{2020}} {2} {Lead research on cloud technologies for heterogeneous control systems}
\project{\ivl{2017}{2022}} {Toma} {Research and applications of hidden aspects in GPU architectures}
\action{\ivl{2017}{2018}} {} {Identified performance imbalances and hidden parallelism in NVIDIA and AMD GPUs}
\action{\ivl{2017}{2018}} {} {Developed method to exploit intra-SM parallelism on NVIDIA (up to 30\% speed-up)}
\action{\ivl{2017}{2018}} {} {Evaluated methods of approximate computing for tomography and accelerated reconstruction
} algorithms by 4-6 times via performance modeling and rebalancing of hardware usage [\urlref{s11554-019-00883-w}{https://doi.org/10.1007/s11554-019-00883-w}]}
%>>>> Better title
\project{\ivl{2012}{2017}} {CMS} {Remote-DMA based Data Acquisition Platform} %Low latency communication for distributed data acquisiton systems}
\action{\ivl{2011}{2013}} {1-2} {Developed a driver platform for fast prototyping of PCIe-based electronics with a
modular (user-space) DMA engine (up to 12 GB/s) and scripting/debugging support}
\action{\ivl{2011}{2013}} {} {Implemented drivers for in-house electronics, e.g. camera (\emph{in use at KIT and Desy})}
\action{\ivl{2011}{2015}} {1-2} {Used RDMA and GPUDirect/DirectGMA to intercommunicate detectors with GPUs}
\action{\ivl{2016}{2017}} {1} {Led a case study on applications of GPUs for CMS track trigger,
decision on acceptance of a track candidate was made within 6~$\mu$s (data transfer + analysis)}
\action{\ivl{2016}{2017}} {1} {GPUs allowed to utilize a more precise algorithm than was possible with classic designs}
\project{\ivl{2010}{2017}} {TPU} {Coordinated student exchange program with Tomsk Polytechnic University}
\project{\ivl{2015}{2016}} {ASTOR/NOVA} {Cloud platform for collaborative analysis of tomographic data}
\action{\ivl{2013}{2018}} {1-2} {Led development of web-visualization for large and time-resolved volumes}
\project{\ivl{2011}{2015}} {UFO} {UFO: Ultrafast tomography with online monitoring and image-based control}
\action{\ivl{2011}{2013}} {1} {Proposed a scalable architecture for pipelined processing of image streams}
\action{\ivl{2010}{2017}} {6} {Coordinated software development with a team in KIT and 3 Russian universities}
\action{\ivl{2013}{2014}} {2} {Supervised development of fast reconstruction algorithms (up to 6 GB/s per node)}
\action{\ivl{2014}{2017}} {1} {Supervised development of regularized reconstruction methods to compensate low SNR and/or undersampling in case of high-speed tomography}
\action{\since{2017}} {} {Since 2016, the developed system is installed at the KIT synchrotron and enables both high-speed
(5 volumes/second) and high-throughput (1000 samples/week) operation}
\project{\ivl{2011}{2014}} {ADEI} {ADEI: Advanced Data Extraction Infrastructure}
\action{2011} {} {Helped to secure funding and initiated collaboration between KIT and YerPhI}
\action{\ivl{2013}{2015}} {4} {Supervised a cross-university team of researchers and engineers}
\action{\ivl{2011}{2014}} {4} {Developed a platform for exploration and analysis of time-series in terascale archives}
\action{\ivl{2014}{2016}} {4} {System is in operation in YerPhI, the SEVAN network, and 7 major facilities at KIT}
\project{\ivl{2009}{2013}} {GPU} {Parallel algorithms and software optimization, 10-100x speed-up}
\action{\ivl{2013}{2014}} {UFO} {Developed parallel algorithms for $\mu$PIV (micro-particle velocimetry)}
\action{2010} {} {Leveraged the PoweXCell architecture for a MRSES feature selection algorithm} %, achieved 5000x speed-up compared to prototype using a single IBM QS22 blade}
\action{\ivl{2009}{2010}} {} {Optimized performance of PyHST (ESRF tomographic framework)}
\action{\ivl{2009}{2010}} {} {Implemented a digital image correlation and tracking algorithm for GPUs}
% \action{\ivl{2007}{2008}} {} {Carried out a performance study of open-source XML frameworks} %(SAX/DOM, Transformation, Schema, Security)
\project{\ivl{2002}{2008}} {DAQ} {Distributed systems for data acquisition and slow control}
% % \action{\ivl{2008}{2009}} {} {Web components to manage and visualize slow control data}
\action{\ivl{2007}{2008}} {} {Stabilized a slow control system of the KATRIN experiment \emph{(in use)}}
\action{\ivl{2005}{2006}} {} {Built a distributed data acquisition system for ASEC particle detector network \emph{(in use)}}
% \action{\ivl{2003}{2005}} {} {Middleware to interconnect heterogeneous components into a unified control system}
\action{\ivl{1999}{2001}} {} {Developed Linux drivers and control software for AI-accelerator}
\end{projects}
|
