Curriculum Vitæ

| iamdeepanshu07@gmail.com

Education

B.Tech in Electronics & Communication Engineering
(July '18 - May '22)
Motilal Nehru National Institute Of Technology
Allahabad, IN
Grade (CPI): 8.68

Experience

QUALCOMM | Associate Engineer
(July '22 – Present)

Hyderabad, IN

  • Engineered Cloud and Edge ML pipelines for on-device AI with sensors stack on Qualcomm’s eAI Runtime Engine, empowering OEMs on Snapdragon 8 Gen 2 and Gen 3 chipsets.
  • Devised a high-precision fixed-point model, achieving 98.9% on-device accuracy for mobile device orientation detection usecase.
  • Contributed in developing Qualcomm’s Low-Power AI (LPAI) DSP package, which enabled Camera, Audio and Sensor ISVs, to develop their AI use cases in Low Power mode.
  • Supported 15+ Qualcomm Global Clients (including Google, Microsoft, Meta and Samsung) with their functional issues on Qualcomm’s Snapdragon Sensors Core Solution, across Mobile and XR Business Units.
Tech Stack:   C/C++ | Python | Pytorch | AIMET

QUALCOMM | Interim Engineering Intern
(May '21 – July '21)

Hyderabad, IN

  • Developed an Auto-Summarizer tool, increasing the Resolution Summary Fulfillment by 95% for all client’s engineering issues on QC SoC, optimizing customer engineer workflows.
  • Implemented a potent Classifier-Decoder Model for precise context extraction from client-CE text interactions, using Advanced Python Scripting and NLP/Language Models for efficient data processing and abstractive summarization.
Tech Stack:   Python | Pytorch | NLTK | Spacy | HuggingFace | T5

Projects

AI-PROCTORING MATE
(Jan '21 – April '21)

💡 : Demo


An examination portal featuring AI-Proctored candidate interface and a Web App for teachers & online quiz creators, facilitating quiz creation, response monitoring, and feedback delivery.

  • Engineered an AI-Based Proctoring System that captures real-time Liveness, Blink, and Gaze from the candidate’s webcam, ensuring thorough monitoring during examinations.
  • Implemented an evaluation backend to evaluate individual candidate responses and instantly update the examiner’s dashboard with relevant quiz statistics, enhancing the real-time assessment process.
  • Implemented Dual User Authentication at Sign Up and Unique Quiz Code Generation on the Quiz Creator’s interface to mitigate unauthorized access.

Tech Stack:  Python | OpenCV | TensorFlow | Django | Tkinter | Selenium ..

Licenses & Certifications

Machine Learning Specialization
(June '23 - July '23)

Coursera
Credential ID:   4WVW8UHYDB95

    Supervised Machine Learning: Regression and Classification
    Credential ID: KNRY37PT82D3
    Unsupervised Learning: Recommenders, Reinforcement Learning
    Credential ID: GDPSMR8CSQX4