Research

novel hardware platforms for AI/ML tasks

In this information age, our computational needs have been growing exponentially. Over the past half-century, these growing computational needs were met by packing a larger number of smaller and faster transistors on a Silicon chip. Now with the transistors close to their smallest possible dimensions, there has been an ongoing effort to explore other alternate technologies. 

• We are working on developing a paradigm to design application specific alternative electronic devices in Silicon to meet these computing requirements in a compact and energy efficient manner.

• The idea of using light to do computation at the speed of light looks more promising than ever with the ongoing developments in Photonic integrated circuits. We are interested in developing an integrated nanophotonic accelerator for accelerating all the linear operations involved in artificial intelligence and machine learning (AI/ML) tasks.

designing robust & Flexible devices

The modern integrated circuits pack billions of electronic devices (transistors) or millions of optoelectronic devices (photodetectors, LEDs, etc.) on a small Silicon chip. One of the challenges in these very large-scale integrated circuits is the inevitable variation in the semiconductor manufacturing process which becomes even more critical for smaller devices. To overcome this issue, we are interested in designing devices that are more robust to process variations.

Many of the principles involved in designing such variation-robust devices also apply to designing devices that are robust to mechanical deformations. This is especially crucial for fabricating optoelectronic and nanophotonic devices on a flexible substrate where mechanical deformations can severely degrade performance.

semiconductor-based photon sources & Detectors

Semiconductors are the workhorse of modern electronic and optoelectronic devices. With the ongoing revolution in quantum technology such as quantum computing, quantum metrology, and quantum communication, there is a heavy thrust to implement these quantum devices on a semiconductor platform for ease of integration. We are actively pursuing semiconductors as an efficient source of few-photon Fock states as well as single-photon detectors which are required for all these quantum applications. Our approach exploits some of the recent developments in 2D semiconductors and photonic metasurfaces to push the frontier of source and detector technology.

scaling limits of photonic devices 

The scaling limit for transistors that power modern electronic devices is now well known. Transistors with active regions comprised of a single atom have been already demonstrated experimentally. Now, with the ongoing advances in the integrated photonic technology, similar scaling limits for the various photonic devices also warrants due attention. A few recent works have indicated that one can design efficient photonic devices comprised of only a few atoms (or a few quantum dots) to manipulate a few photon Fock states. These atomic-photonic devices can not only mimic all the functionalities of classical dielectric-based structures but also offer properties unique to quantum systems such as entanglement. We are actively exploring the design of such atomic-photonic devices and some of the performance tradeoffs involved therein.