Ankush P. Email & Phone Number

California, United States

San Francisco Bay Area

• Architected a scalable API ecosystem: Built microservices infrastructure with Traefik and Kong, supporting GraphQL, REST, and WebSocket protocols. Deployed on Azure with Dockerized services, achieving sub-100ms.
• Developed mobile content syncing tool: Created an on-device OCR tool with YOLO and ONNX, improving model accuracy by 20% and growing datasets by 25% monthly via automated labeling pipelines.
• Engineered task-driven pipelines: Designed pipelines with RabbitMQ, processing 10,000+ scene data points and reducing task execution time by 30%.
• Streamlined annotation workflows: Built tools for data labeling, reducing manual efforts by 50% and accelerating dataset creation for model training.
• Automated ML pipelines: Developed CI/CD workflows with GitHub Actions and MLflow, reducing retraining cycles by 40% and ensuring seamless updates to production models.

San Francisco Bay Area

• Optimized training efficiency for generative models: Integrated stable diffusion models for tasks like text-to-image generation, ControlNet, and LoRA mixins, leveraging distributed training on CUDA/MPS hardware..
• Designed a scalable distributed training system: Architected a distributed system to fine-tune and train PyTorch models across CUDA and MPS hardware, enabling seamless scaling across 100+ nodes.
• Enhanced real-time user experience: Implemented asynchronous updates to optimize real-time image previews and progress tracking, improving user feedback latency by 50%.
• Led full-stack development: Built a robust ecosystem including a React-based frontend for user registration, hardware staking, and real-time progress monitoring, integrated with a Django/GraphQL backend for inference.
• Streamlined backend operations: Orchestrated backend workflows using Celery and RabbitMQ, reducing task processing latency by 35% while supporting billing, email generation, and hardware health checks.
• Enabled community engagement: Designed a frontend platform that showcased over 10,000+ community-generated art pieces, driving user engagement and boosting signups by 25%.

Menlo Park, California, United States

• Built a GraphQL API in Hack/PHP to automate branded content creation
• Led product meetings to build out specifications for branded content platform
• Onboarded teammates onto branded content pipeline
• Designed test suite for ML pipeline

Knoxville, Tennessee Area

• TENNLab - https://neuromorphic.eecs.utk.edu/
• Designed simulations that are used for training and testing neuromorphic algorithms
• Implemented shell scripts for parameter selection in training and for unit testing the framework
• Built a parse engine to convert TENNLab based spiking neural networks (SNNs) into usable networks on Intel's Loihi chip
• Documented software for usage and future development purposes

• CS465/565 - Databases/Scripting Languages (Spring 2020)CS302 - Data Structures & Algorithms II (Fall 2019, Fall 2020)CS140 - Data Structures & Algorithms I (Spring 2020)
• Held weekly office hours to help students with labs and course concepts
• Lead weekly lab sessions to breakdown the labs and offer additional assistance
• Graded homework, labs, and exams in coordination with other teaching assistants

Atlanta, Georgia, United States

• Built a REST API to automate audits:
• Interacted with over 50 million points of data stored in a Cassandra database
• Designed queries to provide audits in near real time
• Developed the API using Java, Spring Boot, and Postman

• Pothole management system used by 30+ employees within Knoxville Public Services
• Co-develop mobile and web applications that automated work orders and routing for potholes
• Organize meetings with clients to generate product deliverables and deliver consumer needs
• Features: routing potholes within the city, generating and managing work orders, and designing a dashboard

Master'S Degree, Computer Science

Bachelor Of Science - Bs, Computer Science And Mathematics

Education record