I work on AI systems that can align with human intent, reason across multimodal information, and behave reliably in real-world settings.
My research sits at the intersection of vision-language models, medical AI, trustworthy machine learning, hallucination evaluation, and agentic AI systems. I am especially interested in building and evaluating AI systems that are not only capable, but also reliable, interpretable, and deployment-ready.
Currently, I work on:
- 🧠 Multimodal AI — vision-language models, medical VQA, image/video understanding
- 🏥 Medical AI — trustworthy VLMs for clinical and biomedical applications
- 🛡️ Trustworthy AI — hallucination detection, uncertainty, safety evaluation, benchmarking
- 🤖 Agentic AI Systems — evaluation, reasoning workflows, and reliable AI agents
- ⚙️ Applied ML Engineering — LLM/VLM fine-tuning, deployment, scalable AI platforms
Beyond research, I have experience leading engineering teams, supervising students, building production ML systems, and collaborating across academia and industry.
Multimodal AI · Vision-Language Models · Medical AI · Trustworthy ML · AI Safety · Hallucination Evaluation · Agentic AI · LLM/VLM Fine-Tuning · Applied Deep Learning
Python · PyTorch · Transformers · vLLM · Hugging Face · OpenAI APIs · Docker · FastAPI · PostgreSQL · Git · LaTeX · Jupyter
- 🔭 Currently working on multimodal, medical, and trustworthy AI systems
- 🌱 Exploring agentic AI evaluation, alignment, reasoning, and safety
- 👯 Open to research collaborations, applied AI projects, and industry-facing AI systems
- 💬 Ask me about VLMs, medical AI, hallucination detection, LLM/VLM fine-tuning, and deployment
- 📫 Reach me through sushant.info.np
- 😄 Pronouns: he/him
- 🌱 Vegetarian
From: 15 June 2026 - To: 22 June 2026
Total Time: 2 hrs 12 mins
Other 17 hrs 15 mins ██████████████████████░░░ 88.63 %
LaTeX 1 hr 9 mins █▒░░░░░░░░░░░░░░░░░░░░░░░ 05.94 %
Documentation 59 mins █▒░░░░░░░░░░░░░░░░░░░░░░░ 05.05 %
HTML+Django 4 mins ░░░░░░░░░░░░░░░░░░░░░░░░░ 00.37 %
