AI Researcher & Developer
Yeabsira
Tesfaye
CS student at Beijing Institute of Technology Focused on NeuroAI and brain-inspired intelligence — currently working on multimodal human state inference and affective computing using physiological and wearable data. Also building multi-agent systems on the side. Seeking AI/ML research internships.
Education
BSc. Computer Science and Technology
Beijing Institute of Technology · Beijing, China
Beijing Municipal Government Scholarship (2024 – Present).
Relevant coursework: Introduction to Artificial Intelligence, Brain Science & Brain-Inspired Intelligence.
Summer School for Generative AI
Tsinghua University · Beijing, China
Lectures: Jie Tang — The Road to AGI; Zhiyuan Liu — LLM Agents; Yuxiao Dong — LLMs and Multimodal Models; Wenguang Chen — Distributed Systems for Large Models.
Experience
AI Researcher
Synheart AI
- Co-authored multimodal Human State Inference research using wearable and mobile sensor data.
- Built syni-cli — a schema-validated LLM framework with persona enforcement, output validation, safety rules, token budget control, and OpenAI / local model support.
- Developed emotion and stress inference pipelines across WESAD, UBFC, MMASH, and DREAMER datasets.
- Implemented HRV-based feature extraction and contributed to affective model training and evaluation.
Software Engineer & Co-founder
Gofere Travels · MMCY Tech · Chiropedic
- 3+ years in web development and software engineering across international companies.
- Co-founded and led a travel-booking and software company — managed operations, development, and team.
Notable Open-Source Contributions
Tencent Cloud · Anthropic
- Youtu-Agent (Tencent Cloud ADP) — Exposed the hidden Memory layer to agents, enabling extended tool capabilities.
- claude-code-security-review (Anthropic) — Fixed an infinite-recursion temp-cleanup bug and added PowerShell-safe tool-call translation.
Publications
Human State Interface Infrastructure
Introduces the SynHeart HSI framework for computing Human State Vectors (HSV) — time-scoped, multidimensional representations of physiological, cognitive, and behavioral states — and defines a canonical interface for exchanging human-state information across AI systems.
Projects
syni-cli
↗Python SDK & CLI for persona-driven, schema-validated LLM interactions. Versioned behavioral contracts, schema-first validation, safety constraints, and token budget tracking.
Pichu
↗Terminal-based coding agent with composable tool stack, sub-agents & task orchestration, MCP ecosystem support, and context / session management.
rubric_ai
↗AI-powered grading platform using LLM agents to evaluate code repositories against instructor-defined rubrics — analysing architecture, documentation, testing, and code quality.
Malaria Diagnosis
↗CNN-based deep learning model to classify malaria-infected patients from microscopic slide images. Trained and evaluated on benchmark biomedical datasets.
Skills
Honors & Awards
Innovation & Entrepreneurship Talent Award
双创英才奖 · Beijing Institute of Technology
Third Prize — Qingju AI Competition
青橘AI "人工智能+" 专项赛
Beijing Municipal Government Scholarship
Beijing Institute of Technology