Srivats Shrikanth

Machine Learning Engineer

Talent

TorontoMember since August 30, 2026

PythonSQLT-SQLPySparkSpark SQLpandasNumPySciPyhypothesis testingregression analysiscorrelation analysisvariance analysisPower BIDAXPower QueryStreamlitPlotlyscikit-learnLightGBMMLflowchronological validationcross-validationROC-AUCPR-AUCBrier ScorePrecision@KGoogle Cloud PlatformDataprocGoogle Cloud StorageSpark DataFrames

Hire this person

Send a job offer directly to this candidate

PythonSQLT-SQLPySparkSpark SQL

About

Analytics and ML professional with a master's degree, a graduate certificate, and more than three years of professional experience, including automation work that reduced monthly reporting effort by approximately 40 hours at a manufacturing company. Currently working as a Machine Learning Engineer at Vosyn Inc., where I build low latency text to speech and voice cloning pipelines on Google Cloud Platform across more than 23 languages. Aim to turn operations into systems that help organizations make decisions.

Experience

Machine Learning Engineer - Vosyn

(2025-12)

Fine tuned multilingual XTTS and FastPitch Text To Speech models and architected GCP Vertex AI/Kubeflow pipelines in Docker, thereby deploying voice cloning and speech synthesis on accelerated GPUs.
Refactored speech synthesis workflows and optimized PyTorch execution using ThreadPoolExecutor parallelization and CPU thread tuning, consequently cutting runtime by 42% without sacrificing audio quality.
Applied W8A32 dynamic quantization, ONNX conversion, and FP16 tuning to VITS architectures, thereby reducing model size by up to 68% and accelerating inference by 30%.
Containerized voice processing pipelines with GCS configurations and automated Cloud Run deployments via Pub/Sub workflows, therefore creating a cloud system that automatically handles data formatting, training, and audio generation.
Engineered mechanisms and fallback logic across OpenVoice, MeloTTS, FastPitch, and Google TTS APIs, as a result establishing voice cloning and speech synthesis coverage for more than 23 languages.

Engineer, Quality - Bonfiglioli Transmissions

(2020-09 - 2023-04)

Automated performance test logging using Python scripts to validate sensor data, thereby reducing processing time by 85% and accelerating critical quality reviews.
Replaced manual QA data entry with automated validation scripts, consequently saving 40 hours of monthly logging overhead and freeing engineers for defect root cause analysis.
Shifted operator recorded Excel inputs into automated sensor data capture workflows, thereby eliminating data entry errors across all test logs.
Developed Power BI dashboards using live sensor data and inspection outputs, as a result providing leadership clear visibility into reliability trends and production risks.