Om Sharma

I’m a software development intern who builds production-minded Android experiences using Kotlin and Jetpack Compose, with an emphasis on reliability, testing, and performance. At Buildlab, I developed an AI outfit recommendation app and integrated 17 REST endpoints with core logic verified using JUnit/Mockito.

I also focus on practical machine learning integration. I fine-tuned a FashionCLIP model on 89K+ image-text pairs and built an automated, gated deployment pipeline for model updates. On-device stability and speed improvements included custom R8 keep rules, fixing Compose recomposition bottlenecks using compiler metrics, and generating Baseline Profiles.

On my main product work, I maintained and shipped Kritique, a faculty-review platform for 4,000+ active students. I owned release management for v3.0.0 (including bookmarking, multi-category reviews, and an edit-review workflow) while mentoring 4 junior developers, and I built an MCP server exposing faculty data to AI clients with tool guardrails verified by a 71-case test suite.

I bring strong observability and systems thinking to my GenAI work by instrumenting MCP server tools with OpenTelemetry for per-tool metrics and distributed tracing, backed by Prometheus and Grafana. I’ve also built Android apps for real-world hackathon-style constraints (e.g., on-device ML with INT8 quantization and multilingual support) and contributed to AutoGPT by implementing a native Slack messaging block with an API client and authentication layer validated by 22 test cases.