Adaickalavan Meiyappan, Ph.D.

- Partner with leadership to define long-term project technical roadmap aligned with business objective. - Build, lead, and mentor a team of high performing engineers and researchers. - Define, scale, and drive the architecture, technical vision, and strategy for the project. - Establish clear project objectives and key results. - Oversee infrastructure and DevOps to optimize performance and reliability. - Engage with broader scientific and engineering community. - Topic: Reinforcement Learning from Human Feedback (RLHF) in Large Language Model (LLM) ◦ Worked on building and training Huawei's Chinese LLM, named Pangu. - Topic: Autonomous Vehicle ◦ Open source development: SMARTS (Scalable Multi-Agent Reinforcement Learning Training School) ◦ Implemented distributed simulation platform for multi-agent reinforcement learning and research on autonomous driving. - Topic: Robotics / Embodied AI ◦ Manipulation of simulated robotic arm. - Topic: Quantum Computing ◦ Quantum algorithms. ◦ Quantum error correction (QEC): Repetition code and surface code ◦ Quantum communication and quantum network explorer. - Topic: Game Theory ◦ Solved game of Hex for board sizes of 9x9 and above. ◦ Implemented depth first search with strategy decomposition in C++. Key idea is to achieve search-space reduction by allowing the losing player to play multiple moves which does not alter the game result.

- Topic: Real-time analytics in video streams ingested from hundreds of cameras. ◦ Tasks include (i) capacitor quality classification, (ii) facial expression recognition, and (iii) fire detection. ◦ Designed deep CNN and stateful-LSTM based machine learning algorithm. ◦ Utilized Google Cloud Engine to run Jupyter Notebook and Python scripts for machine learning. ◦ Extended Python code with C++ libraries, Boost, OpenCV, and CMake. ◦ Accelerated parallel programmming with OpenMP, MPI, OpenCL. ◦ Deployed scalable machine learning models using microservices architecture and Docker containers, orchestrated by Kubernetes. ◦ Developed web API and browser-based real-time streaming visualization for the application. - Topic: Vehicle driver monitoring. ◦ Classifying the cognitive and behavioral distractions of a vehicle driver using physiological signals,driving nature, face, and eye tracking data.

- Designed full digital baseband receiver for near-field communication (NFC) Type-B amplitude-shift keying (ASK) and phase-shift keying (PSK) signals. - Modeled and simulated the entire wireless ASK / PSK transmission, including: Non-return-to-zero (NRZ) ASK / PSK transmitter, inductively coupled NFC-antenna channel, antenna Q-factor, coherent / incoherent detector, ADC, CDR, digital phase-locked loop (PLL), symbol synchronizer, amplitude gain control, adaptive equalizer (FFE & DFE), and symbol detector. - Algorithms considered include: recursive least squares, affine projection, subband adaptive filtering, order-recursive adaptive lattice filter, blind deconvolution. - Translated final production code from floating-point to fixed-point. - Automated chip testing using LabView and ASIC communication using SPI / UART. - Obtained patent on the receiver algorithm in Europe, USA, and China. - Utilized Matlab, Verilog, Cadence Virtuoso, oscilloscope, spectrum analyzer, signal generator, and BERT.

- Fabricated phase change random access memory (PCRAM) cells using novel material compositions and analyzed their read/write performance. - Investigated and suggested suitable reactive ion etch recipes for various phase change materials. - Received the Certificate of Outstanding Achievement.

Supervisors: Prof. Kam Pooi-Yuen & Prof. Kim Hoon Thesis: Digital and Optical Compensation of Signal Impairments for Optical Communication Receivers ◦ Doctoral work focused on L1 physical layer technology, where digital and optical signal processing was applied to the design of fiber-optic communication receivers. Fiber-optic communication was modeled and simulated. ◦ Proposed two novel learning algorithms for impairment compensation and symbol detection, in coherent receivers with linear and nonlinear fiber-optic channels. The algorithms are named CWDAML (complex-weighted decision-aided maximum-likelihood) and ACWDA (adaptive complex-weighted decision-aided). ◦ Demonstrated an optical filter design for simultaneous signal equalization, chirp modulation, and single-sideband filtering, in intensity-modulated direct-detected (IMDD) radio-over-fiber (RoF) receivers, used in wireless optical access networks. The solution proposed is an all-optical-domain signal processing technique.

Minor degree: Optical and Semiconductor Technology Select courses : Modern Optics, Photonics, Digital Signal Processing, Digital Control System, Engineering Electromagnetics, Analog Electronics, Integrated Digital Design, Real-Time Embedded Systems, Microprocessor Systems Select project: Programmed embedded system (microcontroller and FPGA) to build an autonomous terrain navigating vehicle, incorporating real-time communications, sensing, and signal processing. Supervisor: Prof. Yeo Yee Chia Thesis: Diode Switched Crossbar Architecture (1 Diode 1 Resistor) for Ultra High Density Phase Change Random Access Memory

Mathematics (A), Physics (A), Chemistry (A), Biology (A), General Paper (A2), Malay (A1)