Papers by Sanjika Hewavitharana
Mask More and Mask Later: Efficient Pre-training of Masked Language Models by Disentangling the [MASK] Token (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Large-scale pre-trained MLMs can be used to generalize well to a wide range of tasks. |
| Approach: | They propose to append [MASK]s at a later layer to reduce sequence length for earlier layers. |
| Outcome: | The proposed method outperforms RoBERTa for 6 out of 8 GLUE tasks on average by 0.4%. |
Media-to-Insights: A Multi-Agent AI System for Continuous Media Monitoring, Analysis, and Reporting (2026.acl-demo)
Copied to clipboard
Ashraf Hatim Elneima, Ozan Yilmaz, Hadi Nasrallah, Sanjika Hewavitharana, Mohamed Al-Badrashiny, Hassan Sawaf
| Challenge: | Existing systems that use keyword-based media monitoring miss semantically relevant articles and are expensive at scale. |
| Approach: | They propose a multi-agent media monitoring system that processes streaming articles through three stages: article matching, batched feature extraction, and report generation with deterministic deduplication and density-based clustering. |
| Outcome: | The proposed system reduces agent invocations by 20% and reduces core feature extraction calls from 7 to 2 per article - a 71% reduction - with bounded quality tradeoffs . |
Agentic AI for Human Resources: LLM-Driven Candidate Assessment (2026.eacl-demo)
Copied to clipboard
Kamer Ali Yuksel, Abdul Basit Anees, Ashraf Hatim Elneima, Sanjika Hewavitharana, Mohamed Al-Badrashiny, Hassan Sawaf
| Challenge: | Current systems rely on keyword matching and shallow keyword-based screening, leading to missed opportunities and inconsistent evaluations. |
| Approach: | They propose a framework that uses Large Language Models to automate candidate assessment in recruitment. |
| Outcome: | The proposed framework outputs detailed assessment reports, candidate comparisons, and ranked recommendations that are transparent, auditable, and suitable for real-world hiring workflows. |