Help & Supportcareers@gravityer.com
Sr. AI Developer Advocate at Qualcomm | Google Developer Expert in AI and Google Cloud |
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Nano-MoE-JAX
February 24, 2026 – Present
A lightweight, educational Mixture-of-Experts (MoE) GPT-style language model built from scratch in JAX / Flax. Inspired by nanoGPT, NanoMoE replaces the standard FFN in each transformer block with a sparse MoE layer — only top-k experts activate per token, giving increased model capacity.
View ProjectObject-Detection-QNN
December 19, 2025 – December 19, 2025
Object detection (CV). Use a medium-weight model (e.g. YOLOv5/v8 or SSD) converted to ONNX and quantized (INT8). Run it on the Hexagon NPU backend (HTP) using the QNN EP on Windows ARM. Measure throughput and accuracy. This covers quantization and using the NPU for CV detection.
View ProjectEdgeAIApp-ExecuTorch
August 26, 2025 – Present
On-device CLIP inference on Android with ExecuTorch + Qualcomm QNN backend. Zero-shot image classification and vision-language understanding powered by hardware-accelerated AI on Snapdragon devices.
View ProjectJAX-Dataloader
March 19, 2025 – April 15, 2025
A lightweight DataLoader for JAX to load data from various file formats, including CSV, JSON, and more. The goal of this project is to port TensorFlow Dataset (TFDS) functionality into JAX while supporting multiple data sources and preprocessing.
View ProjectCultural Fit Analysis
The candidate's projects are exclusively personal, indicating a strong drive for self-directed learning and exploration in AI/ML. However, the lack of team-based or open-source contributions makes it difficult to assess cultural fit in a collaborative work environment. The projects are highly relevant to a Data Scientist role, especially one with a focus on ML engineering or applied AI.
Soft Skills & Operational Fit
Insufficient data to assess soft skills and operational fit. The candidate's project descriptions are technically focused, but there is no information on collaboration, problem-solving approaches, or communication style in a team setting.