Alvi Chowdhury

Designed subject-specific, research-intensive and adversarial prompts to identify model limitations under complex, out-of-distribution conditions. Performed visual analysis to assess spatial/temporal structure and fidelity across full video sequences and capture subtle degradations.

Trained and optimized advanced NLP models in a virtual machine to improve contextually relevant output generation. Evaluated AI performance for a browser-based virtual assistant, identified failure points, onboarded and mentored team members, and developed prompts with creative teams.

Audited and reconciled inconsistencies in URPC, QIF, and MPC databases to improve product accuracy. Compared multiple data feeds, ensured brand/product alignment, and documented discrepancies.

Trained an autoregressive language model to generate sentences resembling human-like text. Performed annotation based on reinforcement learning from human feedback (RLHF).

Applied data annotation expertise to optimize AI algorithms for operational efficiencies across audio, image, and video platforms for Bangladesh’s largest corporations.

Supervised freelancers and troubleshot drawbacks to support data team deliverables.

Managed HRM system administration and annual evaluation workflows, including PTO count and access system support. Produced analysis and reports and handled asset management for two years, with additional graphic design on select occasions.

Captured and edited video content and supported the director in covering two major events to improve storytelling and production outcomes.

Completed a BSS in Advertising at Independent University, Bangladesh (2017–2026).

Grade: A, B, B

Completed A Levels at Academia School (2013–2016) with grades A, B, B.

Grade: 2A*, 3A, 2B

Completed O Levels at Academia School (2012–2013) with grades 2A*, 3A, and 2B.