Devashish Upadhyay

AI Developer & Data Scientist

Sydney, Australia
(+61) 477130787 | (+91) 9529931971
upadhyaydevashish@outlook.com

Summary

Innovative developer with a proven track record of designing, developing, and deploying end-to-end AI applications to solve complex business challenges. Skilled in building scalable, production-ready solutions using LLM's, NLP, and MLOps. Adept at translating raw data into intelligent systems and actionable insights that drive innovation, efficiency, and measurable outcomes.

Skills

AI & Machine Learning

Python, SQL, Fast API, Machine Learning, Deep Learning, MLOps, Azure AI (Cognitive Services, OpenAI Service, Azure Speech Services), Google AI Services (Vertex AI, Dialog flow), Scikit-learn, TensorFlow, Hugging Face

Cloud Platforms & Computing

Azure, AWS, GCP, Azure Container Apps, Serverless Architecture, Docker, Kubernetes, Azure Functions, Google Cloud Functions

Automation & Operations

DevOps, CI/CD Pipelines (GitHub Actions, Azure DevOps), RPA (Robotic Process Automation), Git, Terraform

Data Engineering & Pipelines

Azure Data Factory, Databricks, Pandas, NumPy, Spark

Data Visualization & Business Intelligence

Tableau, Power BI, Matplotlib, Seaborn, Streamlit

Experience

AI Engineer II, Greenstone Financial Services

May 2025 – Present

Sydney, Australia

Played a pivotal role in creating innovative AI solutions that transformed customer service and operational efficiency. My focus was on developing telephony-based agents and AI-driven quality management systems, which significantly enhanced service delivery. Through automation initiatives, I successfully reduced operational costs and improved overall productivity.

AI Developer, Greenstone Financial Services

Dec 2024 – Apr 2025

Sydney, Australia

As an AI Developer, I design and implement AI-driven telephony solutions to automate and optimize customer interactions. This includes developing outbound lead nurturing agents, automating payment recovery for dishonoured transactions, and creating inbound customer care agents integrated with CRM systems. Additionally, I built an AI-powered quality management system to streamline compliance monitoring and automated the bank reconciliation process to improve operational efficiency and reduce manual effort.

Junior BI Analyst, Greenstone Financial Services

Aug 2023 – Nov 2024

Sydney, Australia

The primary responsibility is to integrate and develop business intelligence data products. This involves tasks such as gathering and analyzing business data to provide valuable insights, collaborating with cross-functional teams to design and implement data models, creating reports and dashboards, and ultimately contributing to a better understanding of Data for the business.

Data Engineer, Geneva Capital Pte

Sep 2021 – April 2023

Singapore

Worked part-time in a fast-paced environment transforming raw data into usable formats, improving data reliability, efficiency, and quality using SQL and Python. Automated analytics solutions and developed interactive metrics and dashboards using Power BI.

Data Science Intern, ACG Worldwide

Jan 2022 – Sep 2022

Mumbai, India

During my internship at ACG, I applied SQL and Python to transform and derive insights from large datasets. Designed, automated analytics solutions, and developed interactive dashboards using Power BI. Worked on advanced statistical methods to support decision making and automation.

Awards

  1. Greenstone Analytics Quarterly Award | Greenstone Financial Services
  2. Greenstone Quarterly Shine Recognition Award | Greenstone Financial Services
  3. Letter of Appreciation | ACG Worldwide

Projects

  1. AI Outbound Lead Nurturing | Greenstone Financial Services

    Designed and deployed AI-powered telephony-based virtual agents capable of executing real-time outbound calls at a scalable rate of 1,000 calls per minute. These agents engaged leads, gauged interest, and performed live warm handovers to contact centre agents for deeper qualification and conversion. The system achieved an average 7-10% transfer rate to humans, significantly enhancing lead engagement efficiency and accelerating the sales pipeline through intelligent automation and seamless human-AI collaboration.

  2. AI Outbound Dishonors | Greenstone Financial Services

    Developed telephony-based AI virtual agents to automate outreach for dishonoured insurance payments. These agents proactively called customers upon payment dishonour events, provided context-aware messaging, and offered seamless transfers to the contact centre for resolution. This solution improved customer contact rates, reduced manual follow-ups, and streamlined the payment recovery process through timely, intelligent intervention.

  3. AI Inbound Customer Care Agents | Greenstone Financial Services

    Developed telephony-based virtual agents integrated with custom CRM to handle tasks like updating contact details, changing addresses, and retrieving policy information, improving efficiency and reducing manual intervention.

  4. AI-Powered Automated Quality Management System | Greenstone Financial Services

    Worked a robust backend framework to evaluate calls systematically against QA compliance standards and industry best practices. This solution automated the quality assessment process, ensuring accuracy, scalability, and consistency in monitoring and reporting. It enables seamless identification of non-compliant instances, providing actionable insights for continuous improvement in call quality and operational efficiency.

  5. Outbound First Payment SMS | Greenstone Financial Services

    Developed an end-to-end system to automate the sending of first payment reminders to customers via SMS. This solution efficiently triggered timely reminders based on customer payment schedules, incorporating context-aware messaging and seamless integration with the backend systems. The system improved payment collection rates, reduced manual intervention, and enhanced the overall customer experience by ensuring prompt communication.

  6. Automated bank reconciliation | Greenstone Financial Services

    Automated the bank reconciliation process, cutting completion time from 10–25 minutes to 30 seconds. Enhanced efficiency and accuracy in financial operations.

  7. Machine Learning and XGboost to predict that covid-19 originated from Bat

    Used genomic sequences of coronavirus to train ML model using XGboost and predict that covid-19 originated from bat.

  8. AI proctored Examination Platform

    Developed AI proctoring system at my own startup Texam for eliminating the need of human proctors which will ultimately save a lot of costs for the institutions while conducting exams.

  9. Machine Learning to predict age and gender of humans (A research project)

    Trained a CNN on datasets like UTK Face, AFAD, and IMDB-WIKI to predict age and gender, achieving high gender prediction accuracy and minimal error in age estimation.

  10. Implementing MLOps using Azure DevOps for several projects | ACG Worldwide

    During the internship, implemented several projects with MLOps Practices (Build CI/CD and Release Pipelines etc.) and industry standards using Azure DevOps, Azure Machine Learning.

  11. Trading Bots (Neural Networks)

    Used Kotak Securities and Binance API to develop automated trading bots and made strategies to trade intraday futures and options.

  12. Convolutional neural network – Hand Gesture Detection using OpenCV

    Trained a CNN using TensorFlow to detect hand gestures using OpenCV to use hardware camera device.

Education

Bachelor of Data Science

2020-23

SP Jain School of Global Management, Sydney, Australia

CGPA 3.3/4

High School

2012-20

St. Paul's Sr. Sec. School, Udaipur, India

12th Grade - 87%; 10th Grade - 85%

Certifications and Trainings

  1. Microsoft Azure AI Engineer (AI-102)
  2. Microsoft Azure Data Analyst (DA-100)
  3. Microsoft Azure Data Scientist (DP-100)
  4. University of Michigan - Python Data Structures (08/2020) Grade Achieved: 98.40%
  5. University of Michigan - Programming for Everybody (08/2020) Grade Achieved: 95.78%
  6. University of Michigan - Retrieving, Processing, and Visualizing Data with Python Grade Achieved: 92.00%