Looking for a job

Parth Vadera

@parthvadera1

Senior AI Engineer at LinkedIn building CareerOS agent workflows and production LLM evaluation pipelines.

At LinkedIn, I designed the CareerOS planner–executor runtime in Python and LangGraph and built agent post-training and evaluation pipelines using Qwen, Flyte, and MLflow.

Previously at TikTok, I developed a real-time deep learning payments fraud model that prevented $12M in fraudulent transactions during its initial month of testing. At Levi Strauss & Co., I led product-level forecasting work that improved MAPE by 26 points over the existing rule-based approach.

Experience

Work history, roles, and key accomplishments

Current

Senior AI Engineer / Machine Learning

Current

Dec 2024 - Present (1 year 10 months)

Designed CareerOS planner–executor runtime in Python and LangGraph, supporting dependency-aware planning and parallel tool execution. Built agent post-training pipelines and LLM-as-a-judge evaluators for iterative quality improvement.

Python LangGraph Qwen LoRAs SFT GRPO Flyte MLFlow VLLM Semantic Retrieval LLM As A Judge Multi Turn Evaluation Tool calling Prompt Evaluation

Machine Learning Lead, Payments

TikTok

Apr 2023 - Dec 2024 (1 year 8 months)

Designed and implemented payments risk strategy architecture and developed a real-time deep learning fraud detection model for financial payments. Led production rollout and replaced a third-party fraud-blocking solution with an in-house model.

Deep Learning Fraud Detection Feature Engineering A B Testing PyTorch TensorFlow Feature Stores Model Monitoring Recommendation Systems Transformer Models Multimodal Learning Attention Mechanisms Contextual Retrieval

Lead Data Scientist

Levi Strauss & Co.

Jan 2019 - Mar 2023 (4 years 2 months)

Led development of global product-level forecasting capabilities and improved MAPE by 26 points. Developed loyalty propensity models and a traffic forecasting engine using Python, XGBoost, Random Forest, and DeepAR.

Python XGBoost Random Forest DeepAR CNN ResNet50 Inception Forecasting Propensity Modeling Recommendation Systems Experimentation A B Testing Pricing Demand Forecasting

Senior Data Scientist

DigitasLBi North America

Aug 2017 - Dec 2018 (1 year 4 months)

Implemented attribution models to measure the impact of TV advertising on web traffic using synthetic control methods, Bayesian structural time series, and LSTM models. Built automation for exploratory data analysis of third-party datasets.

Python R Keras PySpark EC2 Synthetic Control Bayesian Structural Time Series LSTM Attribution Modeling Data Analysis

Data Scientist / Research Assistant

Healthcare Systems Engineering Institute of Northeastern Uni

Jan 2014 - Dec 2016 (2 years 11 months)

Collaborated with medical-record data engineers to build generative classification models using Bayesian belief networks. Led an NSF-funded project to predict referral networks using Bayesian evidence-combination methods.

R SQL scikit learn Networkx Random Forest Kernlab Ggplot2 Bayesian Networks Network Simulation

Data Scientist / Technical Analyst

Tatvic

Jan 2013 - Dec 2013 (11 months)

Developed targeted marketing models for an e-commerce business using customer segmentation, predictive modeling, and A/B testing. Built daily website-visitor forecasting models using Holt-Winters smoothing.

R Google Analytics Customer Segmentation predictive modeling A B Testing Holt Winters Association Rule Mining Feature Engineering