Priya Sharma

Machine Learning Engineer

San Francisco, CA

priya.sharma@email.com

(555) 018-4729

linkedin.com/in/priyasharma-mle

github.com/priyasharma-ml

Skills

Machine Learning & Deep Learning

Frameworks & Libraries

MLOps & Production Engineering

Cloud & Infrastructure

Big Data & Data Processing

Languages & Developer Tools

Education

Master of Science in Computer Science (Specialization: Machine Learning)

Sep 2017 - Jun 2019

University of California, San Diego

Location: San Diego, CA

Bachelor of Technology in Computer Science & Engineering

Aug 2013 - May 2017

National Institute of Technology

Certifications

AWS Certified Machine Learning – Specialty

Jan, 2023

Amazon Web Services

TensorFlow Developer Certificate

Jan, 2020

Google

Summary

Production-focused Machine Learning Engineer with 6+ years of experience designing, training, and deploying scalable ML models and real-time inference systems. Specialized in Deep Learning, NLP/LLM orchestration, recommendation architectures, and MLOps pipelines across cloud and edge platforms. Proven track record of reducing model inference latency by 55%, lowering compute infrastructure costs by $180,000 annually, and serving high-throughput predictive APIs to 10M+ daily active users.

Work experience

Senior Machine Learning Engineer

Apr 2022 - Present

DeepMatrix AI

Location: San Francisco, CA

Designed and deployed a multi-modal recommendation engine using PyTorch and Triton Inference Server, serving 12M+ daily users and driving an 18% lift in platform click-through rate (CTR). Optimized transformer-based NLP and LLM workloads via INT8 quantization, weight pruning, and TensorRT engine compilation, slashing P99 inference latency from 140ms to 42ms. Architected end-to-end automated MLOps pipelines using Kubeflow, MLflow, and Feast Feature Store, accelerating model experimentation-to-production deployment cycles from 4 weeks to 3 days. Built real-time model monitoring and drift detection services using Prometheus, Grafana, and Evidently AI, catching 99% of covariate feature drifts before impacting downstream business metrics. Led distributed multi-GPU training clusters (DeepSpeed, PyTorch FSDP) on AWS EC2 GPU instances, cutting training turnaround times for 7B-parameter models by 45%. Implemented dynamic batching and horizontal autoscaling on Amazon EKS, reducing GPU idle costs by 32% and generating $180,000 in annual AWS savings.

Machine Learning Engineer

Jul 2019 - Mar 2022

Apex Data Labs

Location: Seattle, WA

Built and shipped automated fraud detection classification pipelines processing 3M+ daily financial transactions using XGBoost, LightGBM, and PySpark. Boosted precision-recall AUC by 14% compared to legacy heuristic systems, preventing an estimated $4.2M in annual fraudulent chargebacks. Designed and maintained automated CI/CD pipelines in GitHub Actions to run unit tests, data validation checks (Great Expectations), and model regression benchmarks. Implemented containerized REST and gRPC inference microservices using FastAPI and Docker, maintaining 99.98% uptime at 800 QPS. Collaborated closely with Data Engineering teams to curate high-throughput streaming feature ingestion pipelines utilizing Apache Kafka and Snowflake.

Projects

Domain-Adapted RAG & LLM Agent Engine

Jan 2023 - Dec 2023

Role: Machine Learning Engineer

Built a production Retrieval-Augmented Generation (RAG) agent system using LangChain, Hugging Face transformers, and Milvus vector database for legal document discovery. Implemented hybrid dense-sparse semantic search and cross-encoder re-ranking, boosting retrieval accuracy and answer relevancy by 28%.

Edge Computer Vision Detection Pipeline

Jan 2022 - Dec 2022

Role: Machine Learning Engineer

Developed a lightweight object detection model using YOLOv8 and ONNX Runtime for edge IoT deployment. Reduced model footprint by 65% with zero perceptible loss in mAP@0.5, achieving 45 FPS on edge devices.