Technical Alignment Research Accelerator (TARA)

TARA (Technical Alignment Research Accelerator) is a free 14-week part-time AI safety training program designed for strong Python programmers across the Asia-Pacific region who cannot relocate or commit to full-time programs. It is an initiative of AI Safety Australia & New Zealand (AIS ANZ), with all participant costs covered including compute credits, meals, and study space.

The program is structured around the ARENA curriculum, delivered through full-day in-person Saturday sessions complemented by 2-7 hours of independent study per week. Small groups work through pair programming, covering transformer architectures, mechanistic interpretability, reinforcement learning, and model evaluation techniques. The program concludes with a three-week capstone project where participants can produce publication-quality research with guidance from teaching assistants.

TARA's inaugural cohorts ran in Sydney and Melbourne from March to May 2025, with 19 of 21 participants completing the program (90% completion rate). Within six months of the first cohort, 29% of participants had transitioned into AI safety roles, and two research outputs were published. Building on this success, the March 2026 cohort expanded to six APAC cities — Sydney, Melbourne, Brisbane, Manila, Tokyo, and Singapore — targeting approximately 75-80 participants.

The program is led by Founder and Director Yanni Kyriacos, who previously co-founded AI Safety Australia & New Zealand and spent a decade in marketing and strategy at LinkedIn and News Corp. Technical Program Manager Zac Broeren is a Master of AI student at UNSW and a TARA v1 alumnus. The program's advisory board includes Nelson Gardner-Challis (Technical Lead at Arcadia Impact), Ryan Kidd (Co-Executive Director of MATS), and Dan Wilhelm (visiting researcher at Meridian Cambridge). TARA operates under the fiscal sponsorship of Good Ancestors, an Australian nonprofit.

TARA aims to run two cohorts per year (March-June and September-December), with plans to continue expanding across the Asia-Pacific region to widen the global AI safety talent pipeline.