Samarth Mittal

AI Researcher

San Mateo, California, United States5 yrs 2 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Expert in Machine Learning and Content Understanding.
Proven track record in improving Ads targeting performance.
Developed scalable anomaly detection systems at Nutanix.

Stackforce AI infers this person is a Machine Learning Engineer specializing in SaaS and Computer Vision.

Contact

Skills

Core Skills

Machine LearningContent UnderstandingAds TargetingAnalyticsComputer Vision

Other Skills

PyTorchMachine Learning AlgorithmsData ScienceTime Series ModelingARIMAFB-ProphetSemi-supervised LearningClusteringNatural Language Processing (NLP)Applied MathematicsSQLStatisticsJavapythonDeep Learning

About

My area of interest is LLMs in RecSys. Currently, I am a ML Engineer at Meta working in the Content Understanding and sequential user Modeling. Previously I graduated from UW Madison with masters in CS. I had interned with Meta in the past in the HEC+ team where my job was to improve Ads Targeting for HEC Ad campaigns. Before that I worked for 2 yrs at Nutanix and finished my undergraduate in Computer Science from BITS Pilani, Goa. At Nutanix, I worked on creating a analytics system for network flow logs.Before Nutanix, I was a Research Intern at ROSE Lab, NTU, Singapore there I worked on Object detection on images distorted by different types of noises like Gaussian noise, salt and pepper noise, etc. Before that, I was an intern in the TCS Innovation lab where I worked on the Detection of Handwritten text in Insurance and bank forms.

Experience

Nutanix

Software Engineer

Jul 2019 – Aug 2021 · 2 yrs 1 mo · Bengaluru, Karnataka

Designed and Developed a system for anomaly detection and user behavior entity analytics to monitor a VPC’s network and cloudtrail logs using machine learning and time series modeling methods (ARIMA and FB-Prophet)
Ensured a highly scalable, fault-tolerant, and robust system that could ingest over 10 GBs of data every hour and provide real-time alerts in case an anomaly happens with a False Positive rate of less than 5%.
Implemented a system for grouping VMs and recommending security policies using methods like Agglomerative and spectral clustering along with semi-supervised learning to improve the accuracy of the system by over 15%.

Machine LearningAnalytics

Nanyang technological university

Research Intern

Jan 2019 – Jun 2019 · 5 mos · Singapore

Worked on improving image classification performance of networks like VGGnet and ResNet on images by adding Gaussian noise, salt and pepper noise, and other kinds of artificial noises.
With 2D-Discrete Wavelet Transformation was able to beat the state-of-the-art model with over 10 % accuracy. CIFAR-10 dataset was used for this.

Computer Vision

Tata consultancy services

Research Intern

May 2018 – Jul 2018 · 2 mos · Gurgaon, Haryana

Created a custom CTPN(Connectionist Text Proposal Network) model to beat the state of the art object detection algorithms (YOLOv3, Fast-RCNN) for detecting handwritten text in bank and insurance forms
This was then extended to reading industrial inspection sheets by inferring visual relations and resulted in over 87% accuracy in the text detection process leading to a patent and publication.

Computer Vision