Dhruv AgnihotriLead Software Engineer @ SalesforceAdvanced ML Pipeli|
Dhruv AgnihotriLead Software Engineer @ Salesforce
Advanced ML Pipeli|
>
Personal Projects
TECH SKILLS
HTML5
CSS3
JavaScript
Bootstrap
React JS
Redux
React Router
Material-UI
Next.js
Node.js
Express
Django
Flask
Heroku
AWS
Docker
Kubernetes
Jenkins
TypeScript
Python
Golang
Apex
PostgreSQL
MS-SQL
MongoDBExperience
Salesforce
March 2019 - Present
LMTS Software Engineer & All Star Ranger
Key Responsibilities & Achievements:
- System Architecture & Integration Leadership: Led impactful projects, including engineering a JWT-based authentication system to integrate Sales Cloud and Marketing Cloud, enhancing security and streamlining user access by replacing the legacy refresh token flow. Fostered a culture of innovation and excellence.
- High-Throughput Event Streaming: Designed and implemented a performance-centric, Kafka-based system for streaming email engagement data between Marketing Cloud and Sales Cloud, increasing throughput from 2 billion to 5 billion events per month and drastically cutting latency from 1 hour to approximately 1 minute.
- Automated Package Management: Architected and delivered a containerized Heroku web application for automated MCC package creation and customer push upgrades, integrated with GitHub and Slack for approval workflows, and secured with IAM using SSO.
- Public Cloud Migration (AWS): Spearheaded the migration of core platform services from first-party infrastructure to AWS (Salesforce's Hyperforce), resolving complex architectural challenges and enabling product expansion into GDPR-compliant regions, including India.
University of Michigan
2017 - 2018
Graduate Student Instructor
Key Responsibilities & Achievements:
- Course Instruction: Delivered lectures and facilitated discussions for the CS403 course, covering advanced topics in machine learning and AI.
- Assignment Development: Designed and graded assignments and exams, ensuring they met the highest academic standards.
- Student Support: Provided one-on-one mentoring and group tutoring sessions, enhancing student understanding.
DENSO Automotive - Research Partnership
Dec 2018 - 2020
Human Emotion Classification Researcher
Key Responsibilities & Achievements:
- Deep Learning for Emotion Recognition: Developed a video-based human emotion recognition system using deep neural networks (Inception ResNet in TFlearn). Achieved 69% accuracy in classifying 7 emotion categories, contributing to advancements in human-machine interaction as part of an independent collaborative research project.
- Model Optimization: Utilized FastAI for efficient hyperparameter tuning and model optimization, enhancing the performance and reliability of the emotion detection system.
- Collaborative Research: Engaged in an independent collaborative research partnership with DENSO USA, working with researchers and engineers to drive the project forward.
Shoptelligence
May 2018 - Aug 2018
Data Science Intern
Key Responsibilities & Achievements:
- ETL Pipeline & Data Warehouse Development: Designed and implemented a data warehouse ETL pipeline for high-volume clickstream data, which drove a 30% uplift in user engagement.
- Resilient Data Ingestion: Built a resilient data ingestion system using Apache Pulsar (pub-sub) and CockroachDB for efficient data queuing and transformation.
- Quality Assurance & System Integrity: Implemented over 800 unit tests with Pytest, achieving 95% code coverage, which reduced pipeline failures by 40% and ensured high system integrity.
- Machine Learning API Development: Developed an end-to-end machine learning API for image recognition and catalog item classification, training an Inception V3 model on over 500,000 images.
- Efficient Data Acquisition: Created a Selenium-based Python web scraper with Redis queuing, cutting data acquisition time by 50%.
Altech Infrastructures
June 2015 - Aug 2017
Senior SDE
Key Responsibilities & Achievements:
- Real-Time ETL & Analytics: Developed a real-time ETL pipeline using Spark Streaming and Kafka to analyze industrial boiler sensor data, enabling predictive insights and operational efficiency. This improved data processing speed by 40% and enhanced system reliability.
- Predictive Maintenance Modeling: Improved turbine performance predictions by 20% and Remaining Useful Life (RUL) accuracy by 10% using advanced time-series models and embeddings, reducing downtime by 15%.
- Diagnostic Analytics & Cost Savings: Leveraged R Studio for diagnostic analytics on emission ratios, reducing emission losses by 3% and achieving annual cost savings of $10,000 through data-driven optimizations.
- Optimization Modeling: Enhanced electricity load forecasting using SVM, reducing prediction errors by 3% and optimizing power distribution.