MUFAKIR ANSARI
Data Scientist · Machine Learning Engineer · AI Researcher
39Citations
3h-index
8+Yrs Exp
3.91M.S. GPA

Professional Biography

Mufakir Ansari is a Data Scientist and AI Researcher with more than eight years of combined research and engineering experience building machine learning and data systems across biomedical AI, scientific computing, high-performance computing, and applied analytics. He is currently an AI Researcher whose work focuses on scalable AI systems, intelligent agent design, and scientific workflows.

Mufakir's technical profile is defined by a commitment to rigorous measurement and production-grade delivery. His clinical cancer detection pipeline achieved AUC-ROC 0.950 processing 277,000+ pathology image patches using domain-specific SimCLR pretraining — with false negative rates as low as 0.34%. His Ebola outbreak genomics pipeline automated end-to-end RNA-seq analysis of 356 samples on national HPC infrastructure using SLURM and a fully checkpoint-resumable workflow. His distributional feature selection framework (DDFF) addresses biomarker discovery in high-dimensional, low-sample-size biological datasets where standard methods fail.

Prior to academia, Mufakir served as Technical Lead at Orcinus IT Solutions, delivering end-to-end ML and ETL systems for seven SaaS clients, reducing pipeline latency by 35% and improving deployment reliability by 30%. Earlier, as Business Operations Lead at PanunKart.com, he used data analysis and customer segmentation to support 150% sales growth. His range spans data science, machine learning engineering, analytics, forecasting, HPC systems, NLP, and privacy-preserving product development.


Areas of Expertise

Data Science and Analytics · Machine Learning Engineering · Deep Learning (PyTorch, TensorFlow) · Natural Language Processing · Large Language Models · Computer Vision · Biomedical AI · Scientific Computing · High-Performance Computing · Forecasting and Time Series Analysis · Statistical Modeling · A/B Testing and Causal Inference · MLOps and Reproducible Research · Privacy-Preserving AI · Research-to-Production Delivery


Career History
01/2026 – Present
AI Researcher — Computer Science & Engineering Wright State University · Dayton, OH
Scalable AI systems, scientific workflow automation, intelligent agent design.
06/2024 – 04/2025
Graduate Research Assistant Transportation Systems Research Lab · University of Toledo
Ensemble ML pipelines over 37,000+ records; AWS and Azure Databricks workflows.
08/2023 – 05/2024
M.S. Graduate Research Assistant — HPC Lab University of Toledo · GPA 3.91/4.00
Distributed GPU training optimization; energy-aware SLURM scheduling; doubled training efficiency.
02/2023 – 08/2023
Research Assistant Lamar University · Beaumont, TX
Installation, Testing and Data Analysis for a Weather Station.
05/2018 – 12/2022
Technical Lead Orcinus IT Solutions
End-to-end ML and ETL delivery for 7 SaaS clients; 35% latency reduction; 30% reliability improvement.
09/2017 – 03/2020
Senior Engineer / Consultant MyFajir IT Solutions
Production APIs, BI dashboards, cloud ERP systems; 99.9% uptime SLA.
08/2013 – 09/2017
Business Operations Lead PanunKart.com
Data strategy, segmentation, and experimentation driving 150% sales growth.

Education
08/2023 – 08/2025
M.S., Computer Science & Engineering — AI Track · GPA 3.91/4.00 University of Toledo · Toledo, OH
07/2009 – 07/2013
B.Tech., Electronics & Communication Engineering National Institute of Technology · Srinagar, India

Selected Research & Publications
Under Review — Quantum Machine Intelligence
Dy-Part: A Dynamic, Noise-Aware Scheduler for Optimizing Hybrid Quantum-Classical Algorithms
DOI: 10.21203/rs.3.rs-8041248/v1
Under Review — Signal, Image and Video Processing
High-Sensitivity Detection of Invasive Ductal Carcinoma via Domain-Specific SimCLR Pre-Training
DOI: 10.21203/rs.3.rs-8031909/v1
Preprint
From Text to Returns: Using Large Language Models for Mutual Fund Portfolio Optimization and Risk-Adjusted Allocation
arXiv: 2512.05907v1
Published 2025 — IoTBDS
Enhanced Intrusion Detection in IIoT Networks: A Lightweight Approach with Autoencoder-Based Feature Learning
Proceedings pages 207–214
Published 2025 — American Journal of Computer Science and Technology
Racing to Idle: Energy Efficiency of Matrix Multiplication on Heterogeneous CPU and GPU Architectures

FINAL CREDITS

Let’s build serious AI systems.

Open to relocation. Interested in AI research, applied ML, scientific computing, privacy-preserving agents, and high-leverage product engineering.