• Modeled and delivered fund risk analytics (including risk, liquidity, and econometric models) to support oversight, enhance portfolio construction, and guide investment decisions for emerging medtech startups.

• Leveraged data analytics to improve and innovate fund modeling frameworks, proactively introducing new methodologies to optimize risk assessment and portfolio strategies.

• Utilized causal inference-based econometric models to design and conduct experiments, supporting robust decision science tooling and incrementality testing.

• Drove experimentation and A/B testing methodologies to facilitate rigorous statistical evaluations of MedTech product outcomes and inform strategic investment decisions.

Worked on delivering end-to-end AI and ML solutions for startups, serving as a founding engineer, as well as for large enterprises across fintech, e-commerce, and healthcare. Worked extensively on computer vision, CAD systems, and modern AI technologies including LLMs and retrieval-augmented generation (RAG). Worked on building impactful, real-world AI applications that solved complex problems and drove innovation.

Key projects include:

• Boundary-Aware Object Detection for Diffuse Phenomena (Smog & Flame)

Proposed a deep-learning framework for irregular, low-contrast targets.

Designed an attention-based contour refinement module for arbitrary boundaries.

Planning to develop a GPS-linked forest fire monitoring and notification system using this model.

• Computer-Aided Detection (CAD) for Lung Nodules

Developed a CAD system for lung nodule detection in CT scans using a novel data processing method.

Addressed inconsistent expert annotations in the LIDC dataset to enable robust model training.

Enhanced model sensitivity and interpretability for reliable detection across varied annotation scenarios.

Implemented attention-based deep learning models for precise nodule localization.

Aimed to leverage this method to launch an online lung cancer diagnostic service.

• Object Detection Benchmarking

Compared YOLOv11, Faster R-CNN, and SSD on custom datasets.

Evaluated trade-offs between real-time performance and detection accuracy.

• 2D & 3D Image Generation

Developed a face makeup app allowing control over facial features and face swapping, including face inpainting.

Created 3D human heads from multiple frames and 3D upper-body avatars from a single image, with animation simulation.

• Modern AI & LLM Projects

Built retrieval-augmented generation (RAG) pipelines for knowledge-intensive tasks.

Applied large language models (LLMs) to automate content generation, customer support, and domain-specific Q&A.

• Conducting research in AI, blockchain, digital transformation, and entrepreneurial finance while actively contributing to data science and ML projects.

• Collaborating with interdisciplinary scholars to integrate AI and blockchain technologies.

• Contributing to scholarly publications and presentations at academic conferences and technical forums.