Skip to content
H
Work experience

May 2025 — Jan 2026 · Remote

Algoverse

Machine Learning Researcher

A stateful voice-to-function-calling benchmark spanning 16 service APIs and 456 methods, with an 862-test validation framework.

Function CallingSpeechBenchmarksPython

Highlights

  1. 01

    Engineered a stateful voice-to-function-calling benchmark covering 16 service APIs, 456 methods, 16 backend states, and 7 cross-service scenario sets by building service simulators, executable multistep ground truths, and an 862-test validation framework for function-call syntax, API behavior, and backend-state consistency.

  2. 02

    Assembled a multimodal speech-generation and voice-cloning pipeline across ElevenLabs, Hume AI, gTTS, and VibeVoice, generating 27 demonstrations from 31 prompts and evaluating speech robustness through ambient-audio augmentation and 7 research implementations spanning cloning, synthesis, and voice conversion.