Available for new opportunities

Hi, I'm
Satyam Rai.

Data Scientist &
AI Automation Engineer.

I specialize in building high-throughput data pipelines, optimizing predictive models, and developing deep learning inference systems with C/C++, CUDA, and Python.

Explore My Work GitHub
01.

About Me

Hello! I'm Satyam, a Data Scientist with over 3 years of experience specializing in building and optimizing data pipelines, predictive models, and deep learning inference systems for high-volume datasets.

I have a deep passion for writing highly performant code. I am skilled in Python, SQL, and C/C++, and I excel at developing domain-specific compilers, anomaly detection algorithms, and GPU-accelerated workloads that drastically reduce processing latency.

Beyond the backend engineering, I'm highly experienced in creating interactive data visualizations and dashboards. My goal is always to translate complex, low-level statistical analyses into highly observable, actionable insights for both technical and non-technical stakeholders.

Education

Master of Science, Data Science

University of Maryland, College Park

Aug 2024 - Dec 2025 | GPA: 3.7

Bachelor of Tech, Computer Science

Adamas University

Aug 2017 - Jul 2021 | GPA: 3.9

Tech Stack

Python / C++ / CUDA PyTorch / TensorFlow Apache TVM / Triton GCP / AWS SQL / Dashboards JAX / vLLM
02.

Where I've Worked

AI and Automation Engineer @ Google

May 2025 - Present
  • Innovated new AI systems technologies for efficient LLM inference, using C/C++ and Python to design extensible abstractions.
  • Designed and optimized custom GPU kernels using CUDA C/C++ and Triton, improving throughput for complex reasoning tasks on distributed clusters.
  • Engineered automated evaluation pipelines leveraging PyTorch and JAX to refine safety guardrails and built efficient JIT domain-specific compilers.

Data Scientist @ MSCI

Nov 2023 - Aug 2024
  • Engineered high-throughput inference infrastructure using Apache TVM and ONNX, optimizing financial scoring engines and reducing latency by 28%.
  • Conducted deep performance analysis to identify runtime bottlenecks, utilizing C/C++ and GPU kernel optimization strategies for risk valuation workloads.
  • Made the deep learning stack highly observable by establishing automated documentation and interactive data visualization dashboards.

Data Scientist @ Trove Research

Oct 2021 - Oct 2023
  • Developed foundation models and built anomaly detection algorithms for market telemetry using PyTorch, TensorFlow, and CUDA C++.
  • Optimized deep-learning ingestion pipelines for high-frequency market data using CUDA acceleration, significantly reducing processing latency.
  • Implemented monitoring guardrails for LLM model drift, communicating findings to stakeholders via interactive Plotly visualizations.

Data Scientist @ SHA Infotech

Sep 2020 - Sep 2021
  • Optimized algorithmic efficiency and redesigned database schemas using C/C++ and Python on GCP, cutting query latency by 20%.
  • Implemented automated scraping agents in Python to aggregate competitor data, feeding highly accurate telemetry into a machine-learning analytics platform.
  • Generated SQL-based data visualizations and dashboards that supported strategic decision-making and reduced analysis turnaround time.
03.

Some Things I've Built

Multimodal Video Captioning

Engineered a comprehensive PyTorch evaluation system for generating descriptive video captions, establishing rigorous automated benchmarks to validate semantic alignment across complex datasets.

PyTorch Multimodal AI NLP

Audio-Visual Event Localization

Developed and scaled TensorFlow models for accurate event tagging, implementing advanced multi-modal telemetry analysis to improve localization and ranking metrics.

TensorFlow Computer Vision Telemetry

Contextual Video Summarization

Designed a Transformer-based summarization agent capable of long-form content analysis, leveraging sophisticated attention mechanisms to automatically extract key semantic context.

Transformers LLMs HuggingFace
View More Projects on GitHub
04. What's Next?

Get In Touch

Currently open to new opportunities! Whether you have a question, a potential project, or just want to discuss the latest in AI, my inbox is always open.