We authored a paper on Second Order Jailbreak where we delved into the risks posed by malignant intelligent actors spreading their influence over networks of agents with varying intelligence and motivations.

Key achievements:

• Developed a multi-agent environment for testing jailbreak scenarios (GitHub Repository)
• Created an interactive playground website showcasing agent conversations
• Presented research findings at NeurIPS 2023
• Published our research paper

Co-created and leading a volunteer-driven organization focused on AI safety and alignment, organizing talks, round tables, seminars and bootcamps to foster knowledge exchange and raise awareness about AI risks.

Key achievements:

• Successfully led the AI Safety Fundamentals program from BlueDot Impact, implementing the AI Alignment Curriculum
• Organized a 2-week AI Safety bootcamp (September 2023) for 20 participants
• Delivered in-depth training on:
  • Technical aspects of Transformer architecture
  • Mechanistic interpretability (Induction and Indirect Object Identification circuits)
  • Reinforcement Learning, RLHF
  • LLM jailbreaking techniques and prevention

Working with numerous labs across Switzerland to standardize practices and optimize data management in catalysis research.

Key responsibilities:

• Developing standardized techniques for catalyst storage and sharing
• Facilitating open science through AI model platforms
• Co-authoring research on ML-based detection of metallic atoms in microscope images
• Developing tools for seamless microscope image labeling
• Collaborating with multiple research groups to implement data management best practices

Provided practice sessions to groups of 20-50 students across multiple technical courses.

Courses taught:

• Algorithms and data structures
• Discrete mathematics and probability
• Numerical methods
• Signals and systems
• Python programming

Responsibilities included preparing exercises, conducting practice sessions, grading assignments, and providing one-on-one support to students.

Developed and deployed a production-ready computer vision system for satellite image analysis.

Key achievements:

• Implemented a Mask-RCNN model for:
  • Object detection in satellite imagery
  • Precise segmentation of detected objects
  • Area computation of segmented regions
• Integrated the model with OpenStreetMap data for location-based image matching
• Successfully deployed the model to production environment
• Model is currently being used in production for satellite image analysis

One-year exchange program at EPFL with focus on advanced topics in data science and security.

Key achievements:

• Master thesis: “Point cloud compression for DNA based storage” (Grade: 5.75/6)
  • Built an end-to-end point cloud compressor for quaternary based coding
  • Implemented DNA sequencing, storage, and synthesis constraints

Coursework:

• Cryptography and security
• Information security and privacy
• Decentralized systems
• Advanced topics in machine learning

Extra-curricular:

• Active member of Effective Altruism Lausanne
• Participated in GNU Generation association

Master’s degree in Data Science Engineering with specialization in cryptography and computer systems.

Bachelor’s degree in Engineering with specialization in applied mathematics and computer science.