Summary
Overview
Work History
Education
Skills
Timeline
Generic

Danial Ghasemi

Montreal,QC

Summary

With over seven years of experience working in ML/AI/NLP field and more than ten years of strong academic background in analytical modeling and programming, I strive to bring high-quality intelligence solutions to businesses seeking higher efficiency. Depth knowledge in statistical analysis, optimization, artificial intelligence and strong programming skills, moves my inner motivation to master interesting algorithms in machine learning and data science every single day!

Overview

7
7
years of professional experience

Work History

Senior ML/NLP Scientist and Developer

Cerence Technology
01.2022 - Current
  • Working on fine-tuning a general purpose sentence encoder (or embedding model) for all Cerence NLU needs. This model will be fine-tuned on all Cerence data using contrastive approaches and will be used in our RAG systems.
  • Maintained and developed a modern deep learning codebase involving several tasks meeting Cerence's various model training pipelines. Also, developed comprehensive documentation to facilitate seamless handoff to other users or developers for research and release for production.
  • Developed innovative research methodologies which improved data preprocessing and model convergence in training DNN based models. It improved accuracy and made training pipeline more efficient by +400% relatively.
  • Conducted research and developed transformer based encoder for sequence classification task in our codebase.
  • Performed complex troubleshooting and assisted in leading initiatives to solve challenging customer problems in existing applications. Estimated work hours, divide the work among team members and tracked progress using Scrum methodology on customer bug fixing tickets.
  • Conducted research on improving in-house classifiers and replacing them by more modern DNN models.
  • Developed production ready scripts for trained models to be used in production environments. From data formatting to extraction of model weights from Tensorflow or Pytorch models into a format which is supported for run time application.

ML/NLU Developer

Cerence Technology
09.2019 - 01.2022
  • Developed and collaborated on the next generation of arbitration models which was a more scalable solution in terms of model training and data processing (MACS).
  • Ownership of arbitration project (Mix399) which was a intent classifier on cloud: develop, maintain and improve or adding new features to the existing ML pipeline.
  • Conducted model training, evaluation and integration testing for a cloud project involving over thirty languages globally for Mercedes and BMW, and Audi.
  • Led a project to develop ontology and semantic signatures for NLU models being used at Ford company.
  • Automated repetitive tasks through scripting, freeing up valuable time for higher-priority projects.
  • Coordinated deployments of new software releases, feature updates and fixes. I used to be in a client facing environment and finding innovative ways to fix challenging issues by collaborating with cross-functional teams to ensure seamless integration of new features and debugging complex NLU issues.
  • Mentored junior developers, fostering a collaborative learning environment and supporting professional growth.
  • Adapted quickly to new technologies and programming languages, enhancing overall team productivity.

Data Scientist

Uzinakod
11.2017 - 05.2019
  • Led a project on energy price prediction , I worked on this project end2end from reading white papers on the topic and doing research to develop its ML pipeline and deployment into azure cloud services. I combined deep learning and time series techniques to predict DAP prices in NA markets (NY zone), which led to 90% accuracy!
  • Worked on a face identification project which was aimed to capture people and their identities in real-time in our office space. I used facedetection and OpenCV libraries in the project.
  • Developed a statistical tool for bidding strategy optimization, it was basically a dashboard which was able to optimize different strategies and show their success confidences based on historical data.
  • Cleaned transformed, and analyzed large datasets to uncover hidden trends and patterns for actionable insights.
  • Devised and deployed predictive models using machine learning algorithms to drive business decisions.
  • Collaborated cross-functionally with stakeholders, ensuring alignment on key initiatives and project goals.

Education

Master of Science - Electrical And Computer Engineering

Concordia University
Montreal, QC
01.2018

Bachelor of Science - Electrical And Computer Engineering

Tehran Polytechnic (Amirkabir University)
04.2015

High School Diploma -

Hedayat Nemone
Tehran, Iran
08.2010

Skills

  • Machine learning
  • Deep learning
  • NLP
  • Gen AI
  • Python
  • Pytorch
  • Tensorflow
  • ML pipeline
  • ML deployment
  • Bash (Linux)
  • Computational clusters(Slurm)
  • Hard worker, self-motivated, and self learner
  • Great ability to work independently and within groups

Timeline

Senior ML/NLP Scientist and Developer

Cerence Technology
01.2022 - Current

ML/NLU Developer

Cerence Technology
09.2019 - 01.2022

Data Scientist

Uzinakod
11.2017 - 05.2019

Master of Science - Electrical And Computer Engineering

Concordia University

Bachelor of Science - Electrical And Computer Engineering

Tehran Polytechnic (Amirkabir University)

High School Diploma -

Hedayat Nemone

Similar Profiles

CREATE PROFILE
Danial Ghasemi