Transform Pharmaceuticals: We implemented a system that used AI techniques to automatically design high-throughput experiments from any experiment space specified by scientists. A support vector machine was used to guide the experimentation process. These innovations helped persuade Johnson & Johnson to acquire Transform for $240 million.
Teleron Solutions: We implemented an automatic natural-language question-answering system that was used to provide customer support for Teleron's eCommerce clients.
iCOMS: We implemented a neural-network based system for detecting fraud in credit-card transactions. This system was used to reduce fraud for online stores managed by iCOMS.
Integrated Device Technology: For efficiency, the internal routing of connections within FPGA chips had to be optimized. We implemented a heuristic algorithm to optimize circuits.
Relona: We developed a natural language analysis engine to improve accuracy of web search. To improve CTRs and ad-response rates, we developed an ad-server that analyzed interests of viewers and determined the most appropriate advertisement to display.
Cadence Design Systems: We developed systems to optimize the placement and routing of electronic circuits on boards and within chips.
Mobstac: We designed systems to acquire unstructured data from RSS feeds and display them as structured pages of information on mobiles.
Keisense: We used AI algorithms to implement predictive keyboard input that reduced errors and improved text entry. This innovation helped persuade Nuance Communications to acquire Keisense.