Built RDP with Django, PostgreSQL (RDS) on AWS (EKS), handling 1000+ requests/sec peak for AI researchers, robot app developers, and deployment engineers to access, filter, and modify data via web UIs and Python client

Designed annotation pipelines using RDP, S3, SQS, Lambda functions to send images and videos to vendors so the enriched data can be used for machine learning model training and robot performance metrics

Full list of tech used: Python, Django, pytest/mypy/black, Kubernetes, docker/docker compose, SaltStack, PostgreSQL, Terraform, AWS (EKS, ECR, RDS, SQS, SNS, S3), Datadog (metrics monitoring), Github Actions (CI), Bazel (build system), gRPC/protocol buffers, HTTP, PyTorch, Ray, SkyPilot

Turned the Natural Language Search (NLS) metadata system into a platform to enable self-service NLS support for Salesforce entities with Java and Spring Boot, thus increasing the number of supported entities by 20% in its 1st quarter

Increased number of supported search queries by 20% by implementing NLP tagging algorithms in Java

Enabled users to do query remediation by improving NLP inference logic in Java for a TensorFlow Named Entity Recognition (NER) model to generate multiple interpretations and doing data analysis with Python/Pandas/Matplotlib to select optimal parameters

Used Splunk/Kibana for logs analysis, Grafana for metrics monitoring, and Protocol Buffers (protobufs) for inter-systems communication

Wrote backend code with Java, Spring, JUnit, Mockito and frontend code with JavaScript

Used Splunk for logs analysis

--