About Me

I build agentic systems that bridge the gap between complex data and human efficiency. Currently a Senior Data Scientist at ORO Labs, I have spent over a decade scaling AI— from spearheading GenAI initiatives at Twilio to building foundational speech models at Baidu and Capio. I am a Carnegie Mellon alumnus. I am dedicated to optimizing how deep learning and generative AI models learn and perform in the real world.

Contact Details

Akshay Chandrashekaran
Boston, MA
Email: web@akshayc.com

Download Resume

Education

Carnegie Mellon University, Silicon Valley

Ph.D. Candidate in Electrical and Computer Engineering • January 2012 - December 2019

I was a Ph.D. candidate focused on automated multi-objective hyperparameter optimization for speech recognition. My research included developing hierarchical optimization techniques to jointly improve word error rate and computational efficiency, as well as predicting terminal performance using validation curves from previous configurations. Additionally, I developed speech recognition systems for low-resource languages and implemented LSTM acoustic models for Android using OpenCL.

Carnegie Mellon University

M.S. in Electrical and Computer Engineering • August 2010 - December 2011

Completed M.S. in ECE with a GPA of 3.63/4.00. During this time, I served as a Graduate Assistant researching imagined speech classification using EEG signals and automated axonal bouton detection from visual cortex imagery.

Vishwakarma Institute of Technology

B.E. in Electronics and Telecommunication Engineering • August 2006 - May 2010

Completed B.E. at the University of Pune, India, with a focus on electronics and telecommunications.

Work

ORO Labs

Senior Data Scientist • June 2025 - Present

Incorporating semantic search into supplier selection using agentic frameworks to enable smart, constraint-based querying for purchasing organizations. Improving document feature extraction using multi-modal inference for robust document ingestion.

Floma Inc.

Senior Software Engineer • August 2024 - March 2025

Architected GenAI approaches for marketing campaign messaging and developed creative reasoning frameworks to inject emotion into ad content. Designed transducer-inspired conversational agentic flows and implemented RAG PoCs.

Twilio Inc.

Staff Machine Learning Scientist • June 2019 - August 2024

Spearheaded generative AI initiatives for Twilio Customer AI, including generative audiences and LLM quality control frameworks. Architected ML workflows using Kubeflow and Temporal for speech recognition and custom vocabulary services.

Capio Inc.

Speech Scientist • June 2017 - June 2019

Accelerated optimization of neural network-based ASR models and integrated automated hyper-parameter optimization for model combination and decoders. Developed acoustic and language models for multiple low-resource languages.

Baidu SVAIL

Research Intern • May 2016 - August 2016

Worked on importance sampling-based data sampling techniques to improve training time for speech recognition.

Lenovo

Research Intern • October 2013 - June 2014

Developed a software framework for multimodal interaction on mobile devices, resulting in three patents.

Patents

  1. Multi-modal fusion engine (US Patent 10,649,635), 2020
  2. Selecting multimodal elements (US Patent 10,698,653), 2020
  3. Identification of user input within an application (US Patent 10,613,915), 2020
  4. Efficient feature merging and aggregation for predictive traits (US Patent App. 18/441,140), 2024
  5. Predictive Traits (US Patent App. 18/439,050), 2025

Publications

  1. A. Chandrashekaran, I. Lane, “Auto-ML for Automated Optimization of Speech Recognition on Mobile Devices”, GTC 2018.
  2. K. Han, A. Chandrashekaran, J. Kim, I. Lane, “Densely Connected Networks for Conversational Speech Recognition”, Interspeech 2018.
  3. K. Han, A. Chandrashekaran, J. Kim. I. Lane, “The CAPIO 2017 Conversational Speech Recognition System”, arXiv preprint 1801.00059.
  4. A. Chandrashekaran, I. Lane, “Speeding up Hyper-parameter Optimization by Extrapolation of Learning Curves using Previous Builds”, ECML 2017.
  5. A. Chandrashekaran, I. Lane, “Hierarchical Constrained Bayesian Optimization for Feature, Acoustic Model and Decoder Parameter Optimization”, Interspeech 2017.
  6. A. Chandrashekaran, I. Lane, “Automated optimization of decoder hyper-parameters for online LVCSR”, SLT 2016.
  7. D. Cohen, A. Chandrashekaran, I. Lane, & A. Raux, "The HRI-CMU corpus of situated in-car interactions," IWSDS 2014.
  8. I. Lane, et al., “HRItk: the human-robot interaction ToolKit rapid development of speech-centric interactive systems in ROS,” NAACL-HLT 2012.