ML Engineer & Data Scientist

Rajvrat Thapliyal

Building intelligent systems at the intersection of machine learning, distributed computing, and real-world impact.

Education

Stanford University

Management Science & Engineering

NUS Overseas Entrepreneurship Program

Aug 2025 - Present

National University of Singapore

BSc (Hons) Data Science and Analytics

Special Program in Science (SPS) multidisciplinary research program

Aug 2023 - Present

Experience

Product Manager Intern - AI Applications

Workato Inc · Palo Alto

Aug 2025 - Present
  • Directed the product roadmap for Workato GO, a one-stop AI-powered enterprise productivity platform, defining architecture standards for secure integration of internal systems and external AI agents at scale.
  • Defined and owned the search relevance evaluation strategy, operationalizing NDCG as the primary decision metric and improving search quality by 23% through the implementation of graph database-backed retrieval systems.
  • Architected and deployed automated daily evaluation pipelines on EC2, transforming search quality into a regression-aware, metric-driven shipping gate that directly informed feature rollout decisions.

ML Researcher

Satellite Technology and Research (STAR) Centre

Jan 2025 - May 2025
  • Redesigned physics-informed neural networks for large-scale satellite trajectory propagation across 8000+ Starlink satellites using a 90GB orbital dataset, reducing orbital deviation error by up to 81% across regimes.
  • Refactored and parallelized a monolithic propagator pipeline originally trained on A100 cluster infrastructure, reducing end-to-end training time from 71 days to 13 minutes on a personal GPU via satellite-specific lightweight models.
  • Led architecture design, ablation studies, and error-envelope validation; awarded Best Research Project at NUS SPS Showcase; manuscript currently in preparation for publication.

ML Engineer

OnSchool EdTech Group

May 2024 - Jul 2024
  • Built XGBoost and Prophet ensemble forecasting models and designed a constrained budget allocation optimizer.
  • Increased L1 customer acquisition by 73% within one month by optimizing multi-channel marketing spend.

Projects

Skills

ML & Modeling

PyTorchYOLOv8XGBoostOptuna

GPU & Distributed

TritonSlurmProxmoxDockerKubernetes

Cloud & Infra

AWS EC2/S3Linux

Programming

PythonSQLJavaTypeScript

Scientific Computing

NumPySciPypandasscikit-learn

Contact

Let's work together

If you'd like to discuss a project, research collaboration, or just say hi — I'm always open to connecting.