Open to opportunities

Yaser Keneshloo

@yaserkeneshloo

Message

I build faster, cheaper production LLM, search, and ranking systems.

United States

Message

What I'm looking for

I'm looking to build production LLM, search, ranking, and ML-platform systems where rigorous evaluation, self-hosted models, and measurable improvements in cost, latency, and accuracy matter.

At Wizard Commerce, I lead ML for an AI shopping agent across query understanding, retrieval, ranking, and catalog enrichment. I replaced vendor-model workflows with self-hosted Qwen3 systems, cutting clarifying-question latency from 250ms to 45ms and delivering 96.8% Hit@3 at 8× lower reranking latency.

Previously at IntelyCare and Nomad Health, I built feature stores, automated ML pipelines, personalized recommendation systems, optimization-driven notification and pricing models, and forecasting products. At Fannie Mae, I helped deliver nationwide home-price modeling and financial narrative reports that saved more than $1M annually.

I bring 10+ years of taking ML from research into production, including evaluation harnesses, monitoring, GPU serving, and training infrastructure. Alongside industry work, I've taught applied ML, NLP, deep learning, and LLMs to corporate and U.S. federal engineering audiences for ten years.

Experience

Work history, roles, and key accomplishments

Current

Lead Machine Learning Engineer

Current

Wizard Commerce, Inc.

Feb 2024 - Present (2 years 7 months)

Technical lead for ML behind an AI shopping agent, spanning query understanding, retrieval, ranking, and catalog enrichment. Built batched LoRA adapters, calibrated logprobs, self-hosted cross-encoder reranking, and evaluation harnesses, cutting latency and costs while improving accuracy.

LLM Fine tuning LoRAs Qwen 3 VLLM RAG Cross Encoder Re Ranking Faiss LightGBM LambdaMART BigQuery A B Testing Model Evaluation Harnesses

Current

Contract Instructor, Data Science & Machine Learning

Current

Data Society

Jan 2016 - Present (10 years 8 months)

Designed and taught applied ML curricula in Python, statistics, ML, NLP, deep learning, and LLMs to corporate and U.S. federal engineering audiences.

Python Statistics Machine Learning NLP Deep Learning LLM Teaching

Staff Machine Learning Engineer

IntelyCare

Mar 2023 - Feb 2024 (11 months)

Rebuilt push notifications as an optimization problem with ML relevance and integer linear programming. Stood up Feast feature store, built graph-based job recommendations, and created an internal RAG assistant for analytics questions.

Feast DBT Node2Vec Neo4j LlamaIndex Milvus Cohere LLM Fine tuning Integer Linear Programming RAG

Senior Machine Learning Engineer

Nomad Health

Dec 2021 - Feb 2023 (1 year 2 months)

Introduced Feast as the company's feature store and built automated Kubeflow pipelines. Shipped a job-popularity model and Dynamic Pay Rate optimization, and led graph recommendation and link-prediction initiatives.

Feast Kubeflow Shapley Node2Vec Neo4j GluonTS DeepAR A B Testing Recommendation Systems

Lead Data Scientist

Fannie Mae

Apr 2020 - Dec 2021 (1 year 8 months)

Developed SuperAVM home-price model covering every U.S. property, designed an object-oriented framework for financial narrative reports, and deployed PII detection models scanning millions of records weekly.

PyTorch AutoGluon Sagemaker PII Detection Object Oriented Design AWS

Senior Data Scientist

Marriott International

May 2019 - Apr 2020 (11 months)

Rebuilt the Customer Lifetime Value framework on AWS with XGBoost and neural models, and built the ML behind Global Points Promotion with per-customer cost/benefit models and Gurobi optimization.

XGBoost AWS Gurobi Optimization Customer Lifetime Value Personalization

Machine Learning Engineer

Virginia Tech

Jan 2014 - Dec 2019 (5 years 11 months)

Worked on deep-RL abstractive summarization of legal depositions, prediction of news article click-rate and peak time, and rare-anomaly detection on IoT sensor time series.

Deep reinforcement learning Abstractive Text Summarization Django Docker Time Series Analysis Anomaly Detection

Machine Learning Engineer (Intern)

Verisign, Inc.

Jun 2014 - Aug 2014 (2 months)

Generated valuable new domain names using language models and performed hypergraph community detection for corporate family-tree discovery over 150M+ domain names on Hadoop.

Language Models Hypergraph Community Detection Hadoop Hive MapReduce