M

Mohitdeep Singh

About

Detail

Kano State, Nigeria

Timeline

work
Job
school
Education

Résumé

Jobs verified_user 0% verified
  • Coupang
    Senior AI/ML Engineer
    Coupang
    Feb 2024 - Current (2 years 5 months)
    Built a multi-objective recommender system for Coupang's homepage using hybrid filtering and intent modeling; led to 18% CTR boost.Fine-tuned open-source LLMs (LLaMA-2, Mistral) using product catalog data for a customer support chatbot in EnglishCreated route optimization models for last-mile delivery using dynamic programming and travel-time forecasting.Designed anomaly detection on logistics scan data using deep autoencoders, reducing delivery errors by 22%.Deployed real-time inference APIs on AWS EKS with TorchServe, monitoring via Prometheus + Grafana.
  • Roku Inc
    Senior ML Engineer
    Roku Inc
    May 2022 - Feb 2024 (1 year 10 months)
    Developed predictive models for user retention using XGBoost on daily viewing data, improving churn prediction by 30%.Built NLP-based metadata enhancement pipelines (using spaCy + custom tagging rules) to improve ad targeting on TV content.Led development of lookalike audience models for Roku's DSP using deep autoencoders and K-means clustering.Created automated ML pipeline using GCP Vertex AI for training & deployment of ad ranking models.Partnered with product to design content-based recommendations for Roku Channel using VGG16-based vision features.
  • Pinterest
    Senior Machine Learning Developer
    Pinterest
    Oct 2020 - May 2022 (1 year 8 months)
    Developed visual search models using ResNet50 and feature hashing to match pins with similar visual content.Improved home feed personalization using attention-based ranking models, leading to 22% increase in session duration.Built caption recommendation models using GPT-2 (via Hugging Face 2020 release) fine-tuned on pin metadata.Created automated pipelines in Databricks to process 2B+ daily events for feature engineering.Developed spam and NSFW detection models using vision transformers and content embeddings.
  • Pandora
    Senior Data Scientist
    Pandora
    Dec 2015 - Oct 2020 (4 years 11 months)
    Built collaborative filtering and matrix factorization models (ALS) for radio station personalization.Applied LSTM-based sequence modeling to predict next-song preference based on listening history.Designed genre classification using audio signal processing (Librosa) + early CNN models trained on spectrograms.Created churn models using survival analysis and logistic regression with listener demographics & behavior.Wrote production-grade Spark jobs to process 10B+ song plays/month on Hadoop clusters.
  • Radio
    Data Scientist
    Radio
    May 2014 - Dec 2015 (1 year 8 months)
    Built real-time listener segmentation using K-means clustering on session-level data.Applied CNNs (via Caffe and early TensorFlow) on audio spectrograms for genre classification.Developed rule-based NLP models to tag live shows using TF-IDF + spaCy.Created dashboards to monitor listening trends and peak-time traffic patterns.
  • Intel labs
    Research Intern
    Intel labs
    Dec 2012 - May 2014 (1 year 6 months)
    Researched transfer learning with CNNs for document classification and facial recognition tasks.Built a speech classification model using HMM-DNN hybrid architecture(standard pre-2015 speech technique).Implemented a custom data pipeline to train models on internal video datasets using OpenCV and Theano.Co-authored technical reports and filed internal patent applications on model compression.
Education verified_user 0% verified
  • Carnegie Mellon University
    Master of Science - Robotics & Machine Learning
    Carnegie Mellon University
    Jan 2009 - Jan 2011 (2 years 1 month)
  • Georgia Institute of Technology
    Master of Science - Computer Science
    Georgia Institute of Technology
    Jan 2007 - Jan 2009 (2 years 1 month)
  • Georgia Institute of Technology
    Bachelor of Science - Computer Science
    Georgia Institute of Technology
    Jan 2003 - Jan 2007 (4 years 1 month)