Saksham Thukral

Experience

• 

  The Sparks Foundation

  Oct 2020 - Nov 2020

  Data Science Intern
• 

  Unthinkable Solutions

  Jan 2021 - Jun 2023

  📌Designed and Developed an Automated Speech Recognition engine deployed across 100+ courts in Nigeria.- Worked with Kaldi, ESPnet, Deepspeech, NeMo, and Vosk on Common Voice Dataset for research and POCs.- Automated data preparation to process 8000+ hours of legal domain audio files for distributed model training.- Achieved 90% accuracy in Acoustic and Language models trained with Audio AI and NLP techniques.- Deployed cloud version of ASR as a SaaS for real-time streaming functionality using Websockets, AWS Lambda, ECS, DynamoDB, and Docker.📌Developed scalable Django REST APIs for an E-commerce platform serving over 25+ million customers.- Reduced average page load time from 4.5 to 3 seconds by integrating job queues and message brokers for running asynchronous tasks and implementing caching.- Significantly enhanced code coverage to surpass 90% through unit testing and SonarQube analysis.- Upgraded the CMS of the website with optimized workflows and query optimizations to expedite content updates by 40%.- Performed code integrations via CI/CD pipelines, raised pull requests, and conducted integration testing of APIs.📌AI-Center of Excellence- Collaborated with a team to develop Digital Human technology for healthcare leveraging NVIDIA Audio2Face, NVIDIA Riva, NVIDIA NeMo, and other Generative AI approaches.- Developed a MediaPipe-powered Virtual Try-On feature for a beauty brand with 20+ million users, enabling real-time beauty product application on the face using mobile’s front cameras.- Utilized 3D U-Net for precise volumetric segmentation of MRI images for Brain Tumor detection, enhancing the identification of tissue abnormalities; featured in Software Engineering Insights.- Performed research and developed prototypes of AI solutions for clients from different business domains like Healthcare, E-commerce, and Legal Technologies.- Contributed to defining best practices for the Artificial Intelligence and Data Engineering teams. Show less 📌Developed data pipelines for creating insightful reports- Streamlined merchant reconciliation reports and translated business propositions into actionable insights by developing Looker Studio data analytics dashboards for improved decision-making.- Achieved a 60% reduction in data reporting time by replacing batch data ingestion jobs with streaming data ingestion using GCP Dataflow, Apache Beam, and GCP Pub/Sub.- Improved BigQuery efficiency through query optimization techniques and data partitioning, leading to a 52% reduction in GCP costs and decreased data processing expenditure from 4 GB to 129 MB per refresh.- Implemented hourly data replication from AWS RDS to BigQuery using GCP Datastream.- Created real-time data pipelines for predictive Log Analytics with Kafka ingestion, PySpark processing, Airflow monitoring, and Apache Druid Dashboard.- Reorganized and redesigned Data Warehouse architecture to facilitate near real-time reporting Show less

  • Associate Software Engineer

    Mar 2022 - Jun 2023

  • Jr. Associate Software Engineer

    Jul 2021 - Mar 2022

  • Trainee Machine Learning

    Jan 2021 - Jul 2021
• 

  UWindsor Artificial Intelligence Club

  Oct 2023 - Aug 2024

  Data Science & AI Mentor/Advisor

  As a dedicated Mentor and Advisor at the AI Club of the University of Windsor, my role is centered on providing indispensable guidance and mentorship to club members, facilitating their exploration of AI concepts, projects, and potential career paths. I actively support individuals in overcoming technical challenges by identifying and recommending relevant learning resources and fostering a positive and collaborative learning environment. Additionally, I share valuable industry insights, trends, and best practices, ensuring that our members are well-prepared for the dynamic landscape of the AI field. Show less

• 

  NAV CANADA

  Sept 2024 - now

  Innovative Solutions Developer Co-op